CN102930155B - Obtain the method and device of the early-warning parameters of electricity needs - Google Patents
Obtain the method and device of the early-warning parameters of electricity needs Download PDFInfo
- Publication number
- CN102930155B CN102930155B CN201210425468.6A CN201210425468A CN102930155B CN 102930155 B CN102930155 B CN 102930155B CN 201210425468 A CN201210425468 A CN 201210425468A CN 102930155 B CN102930155 B CN 102930155B
- Authority
- CN
- China
- Prior art keywords
- index
- change
- rate
- leading indicators
- coincidence indicator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 143
- 230000005611 electricity Effects 0.000 title claims abstract description 56
- 239000002131 composite material Substances 0.000 claims abstract description 95
- 230000000737 periodic effect Effects 0.000 claims abstract description 32
- 238000004458 analytical method Methods 0.000 claims abstract description 17
- 238000004364 calculation method Methods 0.000 claims abstract description 14
- 239000000284 extract Substances 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims description 74
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 31
- 238000010606 normalization Methods 0.000 claims description 20
- 230000015572 biosynthetic process Effects 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 6
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 4
- 230000002194 synthesizing effect Effects 0.000 claims description 3
- 230000002596 correlated effect Effects 0.000 claims 1
- 125000004122 cyclic group Chemical group 0.000 abstract description 31
- 238000012216 screening Methods 0.000 abstract description 12
- 230000000694 effects Effects 0.000 abstract description 7
- 230000001932 seasonal effect Effects 0.000 description 18
- 230000001788 irregular Effects 0.000 description 12
- 238000011161 development Methods 0.000 description 9
- 238000005096 rolling process Methods 0.000 description 9
- 239000000203 mixture Substances 0.000 description 8
- 230000003111 delayed effect Effects 0.000 description 6
- 150000001875 compounds Chemical class 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000007774 longterm Effects 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000002459 sustained effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Landscapes
- Electrotherapy Devices (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses the method and device of a kind of early-warning parameters obtaining electricity needs.Wherein, the method includes: obtain the data sequence for generating warning index;Screening acquisition data sequence is carried out according to sequence according to adjusting parameter logistic;Calculate the Trend index of the data sequence including trend term and periodic term, and obtain warning index sequence;Extract the reference index in warning index sequence and selected index;Carry out correlation calculations according to step-out time analysis model and/or K L information computation, to obtain the relative coefficient being chosen between index and reference index, and according to relative coefficient, selected index is screened, to obtain leading indicators and coincidence indicator;Leading indicators and coincidence indicator are synthesized, to obtain the leading composite index number as early-warning parameters and coincident composite Index.By means of the invention it is possible to realize accurately obtaining the early-warning parameters of electricity needs, thus formulate the effect of the counter-measure of reasonable science accurately according to short-term cyclic fluctuation.
Description
Technical field
The present invention relates to power domain, in particular to method and the dress of a kind of early-warning parameters obtaining electricity needs
Put.
Background technology
Economic cyclic swing is the phenomenon of objective reality in socio-economic development, is not with people in process of economic growth
Will is the objective law of transfer, and it is unpractical for attempting to force to eliminate cyclic swing by various artificial means, even at certain
Under the conditions of Xie, artificial pressure elimination cyclic swing also can aggravate fluctuation.By the research to economic cycle moving law, can be
Hold the fluctuation pattern of economic cycle, thus take suitable means to reduce the amplitude of business cycle fluctuation, extend economic fluctuation
Cycle, thus realize the purpose of sustained economic growth.
As a key problem of macroeconomic research, the research of economic cycle is constantly subjected to national governments and numerous
The attention of economist.Along with development and the progress of Econometric technology of economic theory, lot of domestic and foreign scholar starts to use
Economic cycle is monitored and predicts by quantitative method, it is therefore an objective to hold each phase duration of cycle the most exactly
Concrete time that length, turning point occur and expansion and the dynamics etc. of contraction, thus be that government and enterprise are for different cycles
Feature and formation mechenism formulate scientific and rational counter-measure, slow down the amplitude of cyclic swing, reduce cyclic swing and send out economy
The destructiveness that exhibition causes.
Economic activity is the motive force of electricity needs, therefore electricity needs also there will be cycle standing wave move, but electric power need
Asking field, analyzing the means of power demand cycle fluctuation is to contrast economic growth curve for many years and power consumption growth curve,
The integrated economics scholar division to China's stage of economic development, the most subjectively judges that the period of waves of electricity needs is 9-11
Year, the method the most not studying the fluctuation of short term power demand cycle.
In order to solve the problems referred to above, can directly predict the electricity needs growth rate in future by Predicting Technique, analyze it
Fluctuation situation.But, it was predicted that there is a lot of defect in technology:
1, the basic data used without seasonal adjustment, in the Spring Festival, the moon two-day weekend natural law, natural law festivals or holidays, the leap year all
Having a significant impact basic data, the fluctuation tendency predicted based on this will be distortion;
2, no matter use the causality models such as recurrence, per capita household electricity consumption, neutral net, or use ARIMA, logic this
The time series models such as the base of a fruit, are all that the historical trend according to statistical data does reasonably extrapolation, are equivalent to pay close attention to length in Fig. 1
The Changing Pattern of phase trend, can not promptly make answering of the most compound current economic according to the Changing Pattern of long-term trend
To measure.
Short-term cyclic fluctuation distortion, nothing is obtained currently for correlation technique uses Predicting Technique in electricity needs field
Method acquires the early-warning parameters meeting electricity needs, thus causes formulating the rational measure tackling cyclic fluctuation
Problem, the most not yet proposes effective solution.
Summary of the invention
Short-term cyclic fluctuation distortion is obtained, it is impossible to obtain for correlation technique uses Predicting Technique in electricity needs field
Obtain to the early-warning parameters meeting electricity needs, thus cause formulating asking of the rational measure tackling cyclic fluctuation
Topic, the most not yet proposes effective solution, provides a kind of electricity needs that obtains to this end, present invention is primarily targeted at
The method and device of early-warning parameters, to solve the problems referred to above.
To achieve these goals, according to an aspect of the invention, it is provided a kind of early warning obtaining electricity needs is joined
The method of number, the method includes: obtain the data sequence for generating warning index;Carry out according to sequence according to adjusting parameter logistic
Screening, to obtain the data sequence including trend term and periodic term;Calculate the data sequence including trend term and periodic term
Trend index, and according to Trend index, the data sequence including trend term and periodic term is filtered, to obtain early warning
Index series, warning index sequence is that in data sequence, Trend index is the data increased;Extract sending out in warning index sequence
Index on the basis of electricity, and extract the index in addition to generated energy for being chosen index;According to step-out time analysis model and/or K-L
Information computation carries out correlation calculations to warning index, to obtain the dependency between each selected index and reference index
Coefficient, and according to relative coefficient, selected index is screened, to obtain leading indicators and coincidence indicator;Refer to according to synthesis
Leading indicators and coincidence indicator are synthesized by digital-to-analogue type, to obtain the leading composite index number as early-warning parameters and consistent synthesis
Index.
Further, utilize step-out time analysis model and/or K-L information computation that warning index is carried out correlation calculations,
With the relative coefficient between each selected index and reference index, and according to relative coefficient, selected index is sieved
Choosing, includes obtaining the step of leading indicators and coincidence indicator: obtain each selected index according to equation below and benchmark refers to
Relative coefficient between markrL:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) it is quilt
Selective goal, l is the time difference, nlFor the number of all indexs, t=1,2 ..., n is moon number;By time difference value in the first value
In the range of and relative coefficient rlMore than the selected index of first threshold as leading indicators, and time difference value is taken second
In the range of value and relative coefficient rlMore than the selected index of Second Threshold as coincidence indicator.
Further, according to step-out time analysis model and/or K-L information computation, warning index is carried out correlation calculations,
With the relative coefficient between each selected index and reference index, and according to relative coefficient, selected index is sieved
Choosing, includes obtaining the step of leading indicators and coincidence indicator: obtain each selected index according to equation below and benchmark refers to
Relative coefficient r between markl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynOn the basis of)
Index, X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlFor the number of all indexs, t=1,2 ..., n is month
Number;By time difference value in the first span and relative coefficient rlMore than first threshold selected index as in advance refer to
Target Raw performance, and by time difference value in the second span and relative coefficient rlSelected finger more than Second Threshold
It is denoted as the Raw performance for coincidence indicator;To reference index, the Raw performance of leading indicators and the Raw performance of coincidence indicator
It is standardized processing, to obtain standard basis index series pt, standard be chosen sequence q of indext, wherein, standard is chosen
Index includes standard leading indicators and standard coincidence indicator;Obtain each standard as follows and be chosen index and standard base
K-L quantity of information between quasi-indexk: kl=∑ptln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12, T=1,2 ..., n is moon number, and l is the time difference, nlNumber for all indexs;By time difference value in the 3rd value
In the range of and K-L quantity of information klIt is chosen index as leading indicators less than the standard of the 3rd threshold value, and by time difference value
In four spans and K-L quantity of information klLess than the selected index of the 4th threshold value as coincidence indicator.
Further, according to composite index number model, leading indicators and coincidence indicator are synthesized, to obtain as early warning
The leading composite index number of parameter and the step of coincident composite Index include: leading indicators and coincidence indicator carry out symmetrical change respectively
Change processes, to obtain leading indicators symmetry rate of change CW, i(t) and coincidence indicator symmetry rate of change CZ, iT (), wherein, by as follows
Formula carries out symmetrical change process to leading indicators, to obtain leading indicators symmetry rate of change CW, i(t):
Wherein,Be i-th (i=1,2 ..., kw) individual leading indicators, t=2,
3 ..., n, kwNumber for leading indicators;By equation below, coincidence indicator is carried out symmetrical change process, unanimously refer to obtain
The symmetrical rate of change C of markz,i(t):Wherein,Be i-th (i=1,2 ..., kz) individual unanimously
Index, t=2,3 ..., n is moon number, kzIt it is the number of coincidence indicator;To leading indicators symmetry rate of change Cw,iT () is with consistent
Index symmetry rate of change Cz,iT result that () is standardized processing and obtaining after trend adjustment carries out composite calulation, to obtain
Leading composite index number and coincident composite Index.
Further, to leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change Cz,iT () carries out standard
The result obtained after change process and trend adjustment carries out composite calulation, to obtain leading composite index number and coincident composite Index
Step includes: obtain normalization factor A by equation belowW, iAnd AZ, i: t=2,
3,…,n;Use normalization factor Aw,iAnd AZ, iRespectively by leading indicators symmetry rate of change Cw,iT () and coincidence indicator symmetry change
Rate Cz,iT () is standardized processing, to obtain standardization rate of change Sw,i(t) and Sz,i(t), wherein, T=2,3 ..., n;To standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, to obtain
Take standardization average rate of change V of leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorz(t);According to referring in advance
Target standardization average rate of change VwStandardization average rate of change V of (t) and coincidence indicatorzT () carries out composite calulation, to obtain
Leading composite index number Iw(t) and coincident composite Index Iz(t), wherein, And Iw(1)=100, Iz(1)=100。
Further, to standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, with obtain in advance finger
Target standardization average rate of change VwStandardization average rate of change V of (t) and coincidence indicatorzT the step of () including: by as follows
Formula is respectively by the standardization rate of change S of leading indicatorsw,iThe standardization rate of change S of (t) and coincidence indicatorz,iT () is averaged
Rate of change processes, to obtain average rate of change R of leading indicatorswAverage rate of change R of (t) and coincidence indicatorz(t): Wherein, λw,iAnd λz,iIt is that the i-th of leading indicators and coincidence indicator refers to respectively
Target weight;Criterion factor F is obtained by equation beloww:
According to criterion factor FwIt is standardized the average rate of change to process, to obtain the standardization average rate of change of leading indicators
VwStandardization average rate of change V of (t) and coincidence indicatorz(t), wherein, Vw(t)=Rw(t)/Fw, Vz(t)=Rz(t)。
Further, after obtaining the data sequence for generating warning index, method also includes: in data sequence
Data carry out pretreatment, pretreatment includes: fill up missing data process, revise noise data process, data smoothing process with
And data normalization processes.
To achieve these goals, according to an aspect of the invention, it is provided a kind of early warning obtaining electricity needs is joined
The device of number, this device includes: the first acquisition module, for obtaining the data sequence for generating warning index;First processes
Module, for screening according to sequence according to adjustment parameter logistic, to obtain the data sequence including trend term and periodic term;
First computing module, for calculating the Trend index of the data sequence including trend term and periodic term, and according to Trend index
Filtering the data sequence including trend term and periodic term, to obtain warning index sequence, warning index sequence is number
It is the data increased according to Trend index in sequence;First extraction module, is base for extracting the generated energy in warning index sequence
Quasi-index, and extract the index in addition to generated energy for being chosen index;Second computing module, for according to step-out time analysis model
And/or K-L information computation carries out correlation calculations to warning index, to obtain between each selected index and reference index
Relative coefficient, and according to relative coefficient, selected index is screened, to obtain leading indicators and coincidence indicator;The
Two processing modules, for synthesizing leading indicators and coincidence indicator according to composite index number model, to obtain as early warning ginseng
The leading composite index number of number and coincident composite Index.
Further, the second computing module includes: the first sub-computing module, for obtaining each selected according to equation below
Select the relative coefficient r between index and reference indexl:Wherein, l=0, ± 1, ±
2 ..., ± 12, Y=(y1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlFor institute
There is a number of index, t=1,2 ..., n is moon number;First sub-processing module, is used for time difference value in the first span
In and relative coefficient rlMore than the selected index of first threshold as leading indicators, and by time difference value at the second value model
Enclose interior and relative coefficient rlMore than the selected index of Second Threshold as coincidence indicator.
Further, the second computing module includes: the second sub-computing module, for obtaining each selected according to equation below
Select the relative coefficient r between index and reference indexl:Wherein, l=0, ± 1, ±
2 ..., ± 12, Y=(y1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlFor institute
There is a number of index, t=1,2 ..., n is moon number;Second sub-processing module, is used for time difference value in the first span
In and relative coefficient rlMore than the selected index of first threshold as the Raw performance of leading indicators, and time difference value is existed
In second span and relative coefficient rlMore than the selected index of Second Threshold as the Raw performance of coincidence indicator;The
Three sub-processing modules, for carrying out standard to the Raw performance of reference index, the Raw performance of leading indicators and coincidence indicator
Change processes, to obtain standard basis index series pt, standard be chosen sequence q of indext, wherein, standard is chosen index and includes
Standard leading indicators and standard coincidence indicator;3rd sub-computing module, is chosen for obtaining each standard as follows
K-L quantity of information between index and standard basis indexkL:kl=∑ptln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12, T=1,2 ..., n is moon number, and l is the time difference, nlNumber for all indexs;
4th sub-processing module, for by time difference value in the 3rd span and K-L quantity of information klLess than the 3rd threshold
The standard of value is chosen index as leading indicators, and by time difference value in the 4th span and K-L quantity of information klIt is less than
The selected index of the 4th threshold value is as coincidence indicator.
Further, the second processing module includes: the 5th sub-processing module, for leading indicators and coincidence indicator difference
Carry out symmetrical change process, to obtain leading indicators symmetry rate of change CW, i(t) and coincidence indicator symmetry rate of change CZ, i(t), the
Five sub-processing modules include: the 4th sub-computing module, for leading indicators being carried out symmetrical change process by equation below, with
Obtain leading indicators symmetry rate of change Cw,i(t):Wherein,Be i-th (i=1,
2 ..., kw) individual leading indicators, t=2,3 ..., n is moon number, kwNumber for leading indicators;5th sub-computing module, is used for leading to
Cross equation below and coincidence indicator is carried out symmetrical change process, to obtain coincidence indicator symmetry rate of change Cz,i(t):Wherein,Be i-th (i=1,2 ..., kz) individual coincidence indicator, t=2,3 ..., n, kzIt is
The number of coincidence indicator;6th sub-processing module, for leading indicators symmetry rate of change Cw,iT () and coincidence indicator symmetry become
Rate Cz,iT result that () is standardized processing and obtaining after trend adjustment carries out composite calulation, refers to obtain to synthesize in advance
Number and coincident composite Index.
Further, the 6th sub-processing module includes: the 6th sub-computing module, for obtaining standardization by equation below
Factors Aw,iAnd AZ, i: t=2,3,…,n;7th sub-processing module, is used for using mark
Standardization factors Aw,iAnd AZ, iRespectively by leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change Cz,i(t) carry out
Standardization, to obtain standardization rate of change SW, i(t) and Sz,i(t), wherein, t=
2,3,…,n;
8th sub-processing module, for standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, with
Obtain standardization average rate of change V of leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorz(t);7th son meter
Calculate module, for standardization average rate of change V according to leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorz
T () carries out composite calulation, to obtain leading composite index number Iw(t) and coincident composite Index Iz(t), wherein, And Iw(1)=100, Iz(1)=100。
Further, the 8th sub-processing module includes: the 9th sub-processing module, and being used for respectively will be in advance by equation below
The standardization rate of change S of indexw,iThe standardization rate of change S of (t) and coincidence indicatorz,i(t) be averaged rate of change process, to obtain
Take average rate of change R of leading indicatorswAverage rate of change R of (t) and coincidence indicatorz(t):
Further, after performing acquisition module, device also includes: the tenth sub-processing module, for data sequence
In data carry out pretreatment, pretreatment includes: fill up missing data process, revise noise data process, data smoothing process
And data normalization processes.
By the method and device of the early-warning parameters of the acquisition electricity needs of the application, in obtaining original data sequence
After trend term and periodic term, by data sequence screening and calculating are obtained leading indicators and coincidence indicator, then by above-mentioned
Leading indicators and coincidence indicator index synthetic model synthesis obtain early-warning parameters, and analyze electricity needs week according to early-warning parameters
Phase property fluctuates, and solves and uses in electricity needs field Predicting Technique to obtain short-term cyclic fluctuation distortion, nothing in prior art
Method acquires the early-warning parameters meeting electricity needs, thus causes to formulate reasonably reply according to power cycle fluctuation
The measure of cyclic fluctuation, it is achieved that accurately obtain the early-warning parameters of electricity needs, thus accurately according to short-term periodic wave
The effect of the dynamic counter-measure formulating reasonable science, and then slow down the amplitude of cyclic swing, reduce cyclic swing to electric power row
The destructiveness that industry and economic development cause.
Accompanying drawing explanation
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this
Bright schematic description and description is used for explaining the present invention, is not intended that inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is according to the schematic diagram of the cyclic fluctuation of electricity needs in prior art;
Fig. 2 is the structural representation of the device of the early-warning parameters of the acquisition electricity needs according to the present invention;
Fig. 3 is the flow chart of the method for the early-warning parameters of acquisition electricity needs according to embodiments of the present invention;
Fig. 4 is the detail flowchart of the method for the early-warning parameters of acquisition electricity needs according to embodiments of the present invention;
Fig. 5 is the method schematic diagram of the acquisition coincidence indicator according to embodiment illustrated in fig. 4 and leading indicators;And
Fig. 6 is the trend adjustment schematic diagram according to embodiment illustrated in fig. 4.
Detailed description of the invention
It should be noted that in the case of not conflicting, the embodiment in the application and the feature in embodiment can phases
Combination mutually.Describe the present invention below with reference to the accompanying drawings and in conjunction with the embodiments in detail.
Fig. 2 is the structural representation of the device of the early-warning parameters of the acquisition electricity needs according to the present invention.As in figure 2 it is shown,
This device includes: acquisition module 10, for obtaining the data sequence for generating warning index;First processing module 30, is used for
Screen according to sequence according to adjusting parameter logistic, to obtain the data sequence including trend term and periodic term;First calculates
Module 50, for calculating the Trend index of the data sequence including trend term and periodic term, and according to Trend index to comprising
The data sequence having trend term and periodic term filters, and to obtain warning index sequence, warning index sequence is data sequence
Middle Trend index is the data increased;First extraction module 70, refers on the basis of extracting the generated energy in warning index sequence
Mark, and extract the index in addition to power consumption for being chosen index;Second computing module 90, for according to step-out time analysis model
And/or K-L information computation carries out correlation calculations to warning index, to obtain between each selected index and reference index
Relative coefficient, and according to relative coefficient, selected index is screened, to obtain leading indicators and coincidence indicator;The
Two processing modules 110, for synthesizing leading indicators and coincidence indicator according to composite index number model, to obtain as early warning
The leading composite index number of parameter and coincident composite Index.
Use the device of early-warning parameters of the acquisition electricity needs of the present invention, be used for acquisition generating by acquisition module pre-
The data sequence of alert index, then data sequence is screened by the first processing module, and acquisition includes trend term and cycle
The data sequence of item, the first computing module calculates the Trend index of above-mentioned data sequence afterwards, and according to Trend index to above-mentioned
Data sequence filters, and to obtain in data sequence Trend index for the warning index sequence increased, then carries by first
Delivery block extracts index on the basis of the generated energy in warning index sequence, and extracts the index in addition to power consumption for being chosen to refer to
Mark, the second computing module carries out dependency meter according to step-out time analysis model and/or K-L information computation to warning index afterwards
Calculate, to obtain the relative coefficient between each selected index and reference index, and according to relative coefficient to selected finger
Mark screens, and to obtain leading indicators and coincidence indicator, the second last processing module will refer in advance according to composite index number model
Mark and coincidence indicator synthesize, to obtain the leading composite index number as early-warning parameters and coincident composite Index.By this Shen
The device of the early-warning parameters of acquisition electricity needs please, first obtains the trend term in data sequence and periodic term, then to data
Sequence carries out processing acquisition leading indicators and coincidence indicator, and above-mentioned leading indicators and coincidence indicator index synthetic model is closed
Become to obtain early-warning parameters, solve and prior art uses in electricity needs field Predicting Technique obtain short-term cyclic fluctuation mistake
Very, it is impossible to acquire the early-warning parameters meeting electricity needs, thus cause to formulate rationally according to power cycle fluctuation
The problem of counter-measure, it is achieved that accurately obtain the early-warning parameters of electricity needs, thus accurately according to short-term periodic wave
The dynamic effect formulating rational counter-measure, and then slow down the amplitude of cyclic swing, reduce cyclic swing to power industry and
The destructiveness that economic development causes.
Specifically, in electricity needs field, the data sequence for generating warning index includes 4 parts: long-term trend item,
Circulation item that industry expansion-flourishing contractions-decline-further expansion quasi-periodic rule causes, spring, summer, autumn and winter or monthly natural law difference
The random entry that the item in season that causes, ignorance factor cause.By the device of the early-warning parameters of the acquisition electricity needs of the present invention,
First reject the item and random entry in season in original data sequence, obtain the data sequence of trend term therein and periodic term, so
After analyze or K-L quantity of information skill using season GDP or monthly industrial added value as reference index, used time difference correlation
Art screens several leading indicators and coincidence indicator from other economic indicators substantial amounts of, finally uses index synthetic model respectively will
Several leading indicators synthesize beforehand index, several coincidence indicators are synthesized same index, gets early-warning parameters, just
Can realize utilizing the purpose of the future trend of the advanced judgement coincidence indicator of beforehand index according to early-warning parameters.
In the above embodiment of the present invention, data sequence is screened, to initial data by the first processing module 30
Sequence does seasonal adjustment, i.e. seasonally adjusts and the impact of random factor, with long-term trend item (i.e. in above-described embodiment
Trend term) and short-term circulation item (i.e. periodic term in above-described embodiment) based on, synthesized by index screening and index, use scape
The fluctuation tendency of gas early warning technology research electricity needs, and, above-described embodiment economic alarming electricity needs, from substantial amounts of warp
Ji index screens leading indicators and coincidence indicator, is used for judging the cyclic fluctuation of electricity needs so that judged result is more
Accurately so that for formulating the cyclic fluctuation of reasonable counter-measure more accurately rationally.
In the above-mentioned practical work example of the application, the second computing module 90 includes: the first sub-computing module, for according to as follows
Formula obtains the relative coefficient r between each selected index and reference indexl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynOn the basis of)
Index, X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlNumber for all indexs;First sub-processing module, uses
In by time difference value in the first span and relative coefficient rlMore than first threshold selected index as in advance refer to
Mark, and by time difference value in the second span and relative coefficient rlMore than the selected index of Second Threshold as unanimously
Index.
Specifically, model Preliminary screening coincidence indicator is analyzed with reference index as " mark post " screened, used time difference correlation
And leading indicators.The all selected index in addition to the reference index time difference all in advance or in lag period l(i.e. above-described embodiment)
(l=0, ± 1, ± 2 ..., ± 12), the first sub-computing module calculates each selected index and reference index respectively according to following formula
Relative coefficient rl:
Y=(y in above formula1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) for being chosen index, r is phase relation
Number, l represents the advanced or lag period (i.e. the time difference), and l represents advanced when taking negative, represents delayed when taking positive number, l be referred to as the time difference or
Postpone number,WithIt is respectively sequence X and the meansigma methods of Y.nlIt it is the data amount check of all indexs.Then maximum time difference correlation coefficient
Being considered the time difference dependency relation reflecting selected index with reference index, the corresponding number l that postpones represents the advanced or lag period, i.e.
Make time difference relative coefficient rlMaximum delay number l is exactly this selected index and the advanced of reference index or lag period.
According to above-described embodiment of the application, the second computing module 90 can also include: the second sub-computing module, for root
The relative coefficient r between each selected index and reference index is obtained according to equation belowl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynOn the basis of)
Index, X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlNumber for all indexs;Second sub-processing module, uses
In by time difference value in the first span and relative coefficient rlMore than first threshold selected index as in advance refer to
Target Raw performance, and by time difference value in the second span and relative coefficient rlSelected finger more than Second Threshold
It is denoted as the Raw performance for coincidence indicator;3rd sub-processing module, for reference index, the Raw performance of leading indicators and
The Raw performance of coincidence indicator is standardized processing, to obtain standard basis index series pt, standard be chosen the sequence of index
Row qt, wherein, standard is chosen index and includes standard leading indicators and standard coincidence indicator;3rd sub-computing module, is used for
Obtain the K-L quantity of information k that each standard is chosen between index and standard basis index as followsl:
kl=∑ptln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12, T=
1,2 ..., n, l are the time difference, nlNumber for all indexs;4th sub-processing module, is used for time difference value at the 3rd value model
Enclose interior and K-L quantity of information klIt is chosen index as leading indicators less than the standard of the 3rd threshold value, and by time difference value the 4th
In span and K-L quantity of information klLess than the selected index of the 4th threshold value as coincidence indicator.Wherein, the first span
Can be less than-3, the second span may be greater than or equal to-2 and less than or equal to 2, first threshold can be 0.7, the
Two threshold values can also be 0.7.Wherein, t is moon number.
Specifically, relative coefficient r is got by the second sub-computing module calculatinglAfterwards, the second sub-processing module will
Time difference value is in the first span and relative coefficient rlMore than the selected index of first threshold as leading indicators
Raw performance, and by time difference value in the second span and relative coefficient rlMake more than the selected index of Second Threshold
For the Raw performance of coincidence indicator, reference index is done standardization by the 3rd sub-processing module afterwards, the standard base after process
Quasi-index series is designated as pt:
Leading indicators and the coincidence indicator of primary election are also done standardization by the 3rd sub-processing module, the standard quilt after process
The sequence of selective goal is designated as qt:
Then, about the K-L information of reference index after the 4th sub-processing module is calculated as follows each primary election beacon delay l
Amount kl:
kl=∑ptln(pt/qt+1), l=0, ± 1 ..., ± 12
Wherein, l represents the advanced or lag period, and l represents advanced when taking negative, represents delayed, when l is referred to as when taking positive number
Difference, nlBeing the data amount check (numbers of the most all indexs) after data are evened up, the t in above-mentioned formula represents month, and j represents year.
More specifically, after calculating 2L+1 K-L quantity of information, from klValue is selected minima kl′As selected finger
Mark x is about the K-L quantity of information of reference index y, i.e.Its corresponding delay number l*It is exactly that selected index is optimal
Advanced or delayed moon number (season).K-L quantity of information is closer to 0, illustrate index x and reference index y closer to.And will filter out
Leading indicators and the coincidence indicator come are designated as W respectively(3)And Z(3)。
According to above-described embodiment of the application, the second processing module 110 may include that the 5th sub-processing module, for right
Leading indicators and coincidence indicator carry out symmetrical change process respectively, to obtain leading indicators symmetry rate of change Cw,iT () refers to consistent
The symmetrical rate of change C of markZ, iT (), the 5th sub-computing module includes: the 4th sub-computing module, is used for by equation below referring in advance
Mark carries out symmetrical change process, to obtain leading indicators symmetry rate of change Cw,i(t):
Wherein,Be i-th (i=1,2 ..., kw) individual leading indicators, t=2,
3 ..., n, kwThe number of leading indicators;And the 5th sub-computing module, for coincidence indicator being carried out symmetry by equation below
Change process, to obtain coincidence indicator symmetry rate of change CZ, i(t):
Wherein,Be i-th (i=1,2 ..., kz) individual coincidence indicator, t=2,
3 ..., n, kzIt it is the number of coincidence indicator;6th sub-processing module, for leading indicators symmetry rate of change Cw,iT () is with consistent
Index symmetry rate of change Cz,iT result that () is standardized processing and obtaining after trend adjustment carries out composite calulation, to obtain
Leading composite index number and coincident composite Index.
Specifically, the 5th sub-processing module index synthetic model is sought the symmetrical rate of change of index and is processed by the 6th son
Module is by its standardization, by the 4th sub-computing module in the 5th sub-processing module according to following formula pairAsk symmetrical
Rate of change Cw,i(t):
T=2,3 ..., n, wherein,Be i-th (i=1,2 ..., kw) individual elder generation
Row index, kwIt it is the number of leading indicators.
By the 5th sub-computing module according to following formula pairSeek symmetrical rate of change Cz,i(t):
T=2,3 ..., n, wherein,Be i-th (i=1,2 ..., kz) individual one
Cause index, kzIt it is the number of coincidence indicator.
In above-described embodiment of the application, the 6th sub-processing module may include that the 6th sub-computing module, is used for passing through
Equation below obtains normalization factor AW, iAnd AZ, i: t=2,3,…,n;7th son
Processing module, is used for using normalization factor AW, iAnd AZ, iRespectively by leading indicators symmetry rate of change Cw,i(t) and coincidence indicator pair
Claim rate of change Cz,iT () is standardized processing, to obtain standardization rate of change Sw,i(t) and Sz,i(t), wherein,
8th sub-processing module, for standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, with
Obtain standardization average rate of change V of leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorz(t);7th son meter
Calculate module, for standardization average rate of change V according to leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorz
T () carries out composite calulation, to obtain leading composite index number Iw(t) and coincident composite Index Iz(t), wherein, And Iw(1)=100, Iz(1)=100。
Specifically, the 6th sub-computing module according to equation below to leading indicators symmetry rate of change Cw,i(t) and coincidence indicator
Symmetrical rate of change Cz,iT () does standardized calculation so that it is average absolute value is equal to 1, obtains normalization factor Aw,iAnd AZ, i:
Then the 7th sub-processing module is according to Aw,iAnd AZ, iRespectively by Cw,i(t) and Cz,iT () standardization, obtains standardization and becomes
Rate Sw,i(t) and Sz,i(t):
T=2,3 ..., n, the 8th sub-processing module changes according to standardization afterwards
Rate Sw,i(t) and Sz,i(t) be averaged rate of change process, to obtain standardization average rate of change V of leading indicatorsw(t) and one
Cause standardization average rate of change V of indexz(t), and averagely become according to the standardization of leading indicators by the 7th sub-computing module
Rate VwStandardization average rate of change V of (t) and coincidence indicatorzT () carries out composite calulation, to obtain leading composite index number Iw(t)
With coincident composite Index Iz(t), specifically: make Iw(1)=100, Iz(1)=100, then
More specifically, getting leading composite index number Iw(t) and coincident composite Index IzAfter (t), electric power can be needed
Asking trend to be adjusted, method is as follows:
According to following compound interest formula each sequence of coincidence indicator group obtained respectively respective average rate of increase:
Wherein, Fig. 6 is the trend adjustment schematic diagram according to embodiment illustrated in fig. 4, and as shown in Figure 6, t is month,WithIt is that coincidence indicator group i-th index circulates with last at first respectively
Meansigma methods, mIiWith mLiBe respectively coincidence indicator group i-th index at first with the most metacyclic moon number, k2It it is coincidence indicator
Number, miIt is the center circulated at first to the moon number between the most metacyclic center.
Then the average rate of increase G of coincidence indicator group is obtainedr, and as target trend:Afterwards
To the leading and initial composite index number I of coincidence indicatorw(t) and IzT () obtains their respective balanced growth with compound interest formula respectively
Rate r 'wWith r 'z:
Wherein,
Standardization average rate of change V to leading indicators group and coincidence indicator group againw(t) and VzT () does trend adjustment:
V′w(t)=Vw(t)+(Gr-r′w), V 'z(t)=Vz(t)+(Gr-r′z), t=2,3 ..., n.Then according to above-mentioned enforcement
Method in example calculates composite index number: make I 'w(1)=100, I 'z(1)=100, then
Generate with leading composite index number CI that the benchmark time is 100w(t) and coincident composite Index CIz(t):
WhereinWithIt is I ' respectivelyw(t) and I 'zT () is in the meansigma methods in benchmark time.
According to above-described embodiment of the application, the 8th sub-processing module may include that the 9th sub-processing module, is used for passing through
Equation below is respectively by the standardization rate of change S of leading indicatorsw,iThe standardization rate of change S of (t) and coincidence indicatorz,i(t) carry out
The average rate of change processes, to obtain average rate of change R of leading indicatorswAverage rate of change R of (t) and coincidence indicatorz(t):
9th sub-computing module, for according to criterion factor FwIt is standardized the average rate of change to process, to obtain
Take standardization average rate of change V of leading indicatorswStandardization average rate of change V of (t) and coincidence indicator2(t), wherein, Vw(t)
=Rw(t)/Fw, Vz(t)=Rz(t)。
Specifically, by equation below respectively by the standardization rate of change S of leading indicatorsw,iThe standard of (t) and coincidence indicator
Change rate of change Sz,i(t) be averaged rate of change process, to obtain average rate of change R of leading indicatorsw(t) and coincidence indicator
Average rate of change Rz(t):
Then according to equation below gauge index normalization factor Fw:
Finally according to criterion factor FwIt is standardized the average rate of change to process, to obtain the standard of leading indicators
Change average rate of change VwStandardization average rate of change V of (t) and coincidence indicatorz(t):
Vw(t)=Rw(t)/Fw, Vz(t)=Rz(t), t=2,3 ..., n, wherein, by the mean change of coincidence indicator sequence
The amplitude of rate goes to adjust leading indicators sequence and the average rate of change of lagging indicator sequence, its purpose is to two indexes to work as
Make a harmonious system to apply.
According to above-described embodiment of the application, after performing acquisition module 10, device can also include: the tenth son processes
Module, for the data in data sequence are carried out pretreatment, pretreatment includes: fills up missing data and processes, revises noise number
Process according to process, data smoothing and data normalization processes.
Fig. 3 is the flow chart of the method for the early-warning parameters of acquisition electricity needs according to embodiments of the present invention.Fig. 4 is basis
The detail flowchart of the method for the early-warning parameters of the acquisition electricity needs of the embodiment of the present invention.
As shown in Figure 3 and Figure 4, the method comprises the steps:
Step S102, obtains the data sequence for generating warning index.
Wherein, this step can be realized by step S202 in Fig. 4: collects macroeconomy and the electricity needs moon number of degrees
According to, and do pretreatment.
Step S104, screens according to sequence according to adjusting parameter logistic, includes trend term and periodic term to obtain
Data sequence.Wherein, the method can being realized by step S204 in Fig. 4: by data sequence being made seasonal adjustment, obtaining institute
There are trend term and the periodic term of index.
Specifically, in electricity needs field, the data sequence generating warning index includes 4 parts: long-term trend item (i.e. becomes
Gesture item), industry expansion-flourishing contraction-decline-further expansion quasi-periodic rule cause circulation item (i.e. periodic term), the spring and summer autumn
Data are the magnitude of value and physical quantity by the random entry that winter or item in season that monthly natural law difference causes, ignorance factor cause
Index makees seasonal adjustment, rejects the item and random entry in season in data sequence, obtains the data of trend term therein and periodic term
Sequence.
Step S106, calculates the Trend index of the data sequence including trend term and periodic term, and according to Trend index
Filtering the data sequence including trend term and periodic term, to obtain warning index sequence, warning index sequence is number
It is the data increased according to Trend index in sequence.Wherein, this step can be realized by step S206 to S208: step S206 is sentenced
Whether the achievement data in disconnected data sequence is growth indices class achievement data, and where it has, perform step S208:
The growth rate of parameter, in the case of the achievement data in data sequence is not growth indices class achievement data, performs
Step S210.
Step S108, extracts index on the basis of the generated energy in warning index sequence, and extracts the finger in addition to generated energy
It is designated as being chosen index.Wherein, the method is realized by step S210: determine reference index.
Specifically in electricity needs field, using season GDP or monthly industrial added value as reference index.
Step S110, carries out correlation calculations according to step-out time analysis model and/or K-L information computation to warning index,
To obtain the relative coefficient between each selected index and reference index, and according to relative coefficient, selected index is entered
Row filter, to obtain leading indicators and coincidence indicator.Wherein, step S212 can realize the method: sieves with step-out time analysis model
Select initial coincidence indicator and initial leading indicators;Then step S214 is performed: by the K-L information computation coincidence indicator from primary election
With leading indicators screens final coincidence indicator and leading indicators.
Specifically, can be consistent by step-out time analysis model Preliminary screening from the index such as macroeconomy, the output of industrial product
Index and leading indicators.
Step S112, synthesizes leading indicators and coincidence indicator according to composite index number model, to obtain as early warning
The leading composite index number of parameter and coincident composite Index.Wherein, said method is realized by execution step S216: synthesize with index
Model synthesizes same index and beforehand index coincidence indicator and leading indicators respectively;Then step S218 is performed: according to one
Cause index and beforehand index obtains early-warning parameters, and use early-warning parameters to analyze power demand cycle fluctuation.
Specifically, respectively several leading indicators are synthesized beforehand index with index synthetic model, several are consistent
Index synthesizes same index, gets early-warning parameters, it is possible to realize utilizing the advanced of beforehand index according to early-warning parameters
Judge the purpose of the future trend of coincidence indicator.
The method using the early-warning parameters of the acquisition electricity needs of the present invention, by being used for generating warning index by acquisition
Data sequence screens, and obtains the data sequence including trend term and periodic term, calculates above-mentioned data sequence afterwards
Trend index, and filtering this data sequence according to Trend index, to obtain Trend index in data sequence for increasing
Warning index sequence, then extracts index on the basis of the generated energy in warning index sequence, and extracts the finger in addition to power consumption
It is designated as being chosen index, according to step-out time analysis model and/or K-L information computation, warning index is carried out dependency meter afterwards
Calculate, to obtain the relative coefficient between each selected index and reference index, and according to relative coefficient to selected finger
Mark screens, to obtain leading indicators and coincidence indicator, finally according to composite index number model by leading indicators and coincidence indicator
Synthesize, to obtain the leading composite index number as early-warning parameters and coincident composite Index.By the acquisition electric power of the application
The method of the early-warning parameters of demand, is first obtained the trend term in data sequence and periodic term, is then obtained by screening and refer in advance
Mark and coincidence indicator, and above-mentioned leading indicators and coincidence indicator index synthetic model are synthesized acquisition early-warning parameters, solve
Prior art use Predicting Technique obtain short-term cyclic fluctuation distortion in electricity needs field, it is impossible to acquire and meet electricity
The early-warning parameters of power demand, thus the problem causing to formulate rational counter-measure according to power cycle fluctuation, it is achieved
The early-warning parameters of accurate acquisition electricity needs, thus formulate according to short-term cyclic fluctuation accurately and reasonably tackle periodically
The effect of the measure of fluctuation, and then slow down the amplitude of cyclic swing, reduce cyclic swing and power industry and economic development are made
The destructiveness become.
In the above embodiment of the present invention, first data sequence is screened by step S102, namely to original number
Do seasonal adjustment according to sequence, i.e. seasonally adjust and the impact of random factor, with long-term trend item (i.e. in above-described embodiment
Trend term) and short-term circulation item (i.e. periodic term in above-described embodiment) based on, synthesized by index screening and index, use
The fluctuation tendency of prosperity early warning technical research electricity needs, and, above-described embodiment economic alarming electricity needs, from substantial amounts of
Economic indicator is screened leading indicators and coincidence indicator, is used for judging the cyclic fluctuation of electricity needs so that judged result is more
Add accurately so that for formulating the cyclic fluctuation of reasonable counter-measure more accurately rationally.
Wherein, the pretreatment in step S202 may include that filling up missing data processes, revises noise data process, number
Process according to smoothing processing and data normalization, and the data sequence after processing is as Y(0)。
Specifically, in above-described embodiment of the application, before performing step S204, the method also includes walking as follows
Rapid:
(1) festivals or holidays are removed or given month that other reason causes working days are how many between different year difference
Not, concrete grammar is as follows:
In this embodiment, if monthly data sequence to be analyzed(t=1,2 ..., n) total m, n month, n=m ×
12, t represent moon number, and i represents number of weeks, and j represents year number, and the most each moon real work natural law is Dt(t=1,2 ..., n), m puts down
The work natural law in equal each month is:(L=1,2 ..., 12), work natural law can be obtained and adjust system
Number Sequence pt:
T=1,2 ..., n, then according to work natural law regulation coefficient ptObtain the moon number of degrees after adjusting in month
According to sequence Y(1):
(2) the inside of a week in sequence is adjusted: from former sequence, extract the inside of a week (week structure) because of each moon out no
The variation caused together.
Assuming that the inside of a week variable factor is included in irregular key element, the form of i.e. irregular key element is IDr, it is assumed that
ID is decomposited from former sequencer, obtain Monday with regression analysis, two ..., the respective weights of day, thus by IDrIt is decomposed into true
Positive irregular key element I and the inside of a week variable factor Dr。
In above formula: IDrtFor including t month the inside of a week variable factor DrIrregular key element;xitFor week i in the t month
Natural law (t=1 ..., n);BiFor week i weight ();AtFor the natural law of the t month, take 28.25 days February;ItFor really
Irregular key element.
At thus obtained BiEstimated value be biTime, the inside of a week variable factor of the t month can be calculated according to equation below
Dr:
Drt={x1t(b1+1)+x2t(b2+1)+…+x7t(b7+1)}/At。
(3) after data sequence being adjusted according to festivals or holidays and the inside of a week factor, the special item in sequence is entered
Row is revised.
When decomposing the various factors in economic time series, need to revise in advance have in erratic variation the most different
The item (the most special item, such as strike, the impact of awful weather, error in data etc.) of constant value.Its method is:
The setting of the boundary value of the most special item.
Assume to decomposite from former sequence irregular key element I.In order to get rid of the exceptional value in Irregular variation I,
Need to calculate 5 years rolling average standard deviations of I.First initial 5 year rolling average standard deviation is calculatedThat is:
In above formulaBeing 5 years moving averages of I sequence, m is the year number of I sequence, t=1,2 ..., n(t is the moon of I sequence
Number).AllowCorresponding to the center year during 5 years, calculate one every yearThereforeIt it is year degree series.It is believed that it is full
FootItIt is special, removes these It, by following formula:
Again calculate 5 years mobile standard deviation { σj, t=1 in formula, 2 ..., n(t is the moon number of I sequence), a is special datum
Number.{σjSequence two ends respectively lack 2, are respectively adopted distance top and the terminal { σ of the 3rd yearjReplace that two ends are short of two
The σ in yearjValue.
The correction of the most special item.
According to 5 years mobile standard deviation { σjCalculate flexible strategy w of correction,
Wherein, t=1,2 ..., n(t is the moon number of I sequence), j=
1,2 ..., m, utilize above-mentioned inequality can revise the special item of I sequence: corresponding to wt< It of 1, with this wtFor flexible strategy, therewith phase
The I of each two before and after Jint-2,It-1, It+1, It+2(notice that the w corresponding to item taken is necessarily equal to 1, otherwise take side
Value) totally 5 make weighted average, by the value so obtainedReplace It.If corresponding to wt< the I of 1tWhen being positioned at two ends, with this wtFor
Power, 3 ws close with ittThe I value of=1 totally 4 make weighted average, by this meansigma methods obtainedReplace It.Revise special item
After I sequence be designated as Iw。
(4) according to monthly data sequenceCarry out initial estimation, obtain initial trend circulating component.
Use the trend circulating component in 12 rolling average estimated sequences of centralization
Wherein, yt-6, yt-5..., yt,…,yt+5,yt+6It it is monthly data sequenceIn element.
(5) according to trend circulating componentEstimate irregular composition in season
(6) according to irregular composition in seasonTo application of each month 3 × 3 rolling averages composition in season according to a preliminary estimate:
First according to equation below, seasonal factor is obtained
Then it is standardized seasonal factor calculating, obtains standard seasonal factorSo that factor sum is each
Individual continuous print is all approximately zero in 12 months:
(7) according to standard seasonal factorSequence after first estimation seasonal adjustment
(8) trend circulating component is estimated further with 13 rolling averages
(9) irregular composition in season is estimated further:
(10) with the composition in season that 3 × 5 rolling averages estimations are final:
Final seasonal factor is obtained according to equation below
Then to seasonal factorIt is standardized obtaining standardized final seasonal factorFactor sum is existed
Each continuous print is approximately zero in 12 months:
(11) according to standardized final seasonal factorEstimate the sequence after seasonal adjustment
(12) obtaining final trend circulating component, its method is as follows:
First irregular composition is estimated:
Then irregular item is usedAnd trend termThe moon rate of increase the ratio of absolute value sum weigh irregularly
The significance of composition
IfThen with the trend circulating component that following 9 rolling averages estimation is final:
IfThen with the trend circulating component that following 13 rolling averages estimation is final:
IfThen with the trend circulating component that following 23 rolling averages estimation is final:
Final trend circulating component is got through above-mentioned calculating
(13) according to trend circulating componentWith the sequence after seasonal adjustmentEstimate final irregular composition:
Then get sequence Y only comprising trend circulation item after seasonal adjustment(2)=TC(3)。
In above-described embodiment of the application, in Fig. 4, step S108 realizes especially by following method: to physical quantity and valency
Value figureofmerit, the development index of the sequence after calculating seasonal adjustment: It is the index of development index,
Do not convert, i.e. Y(3)=Y(2)。
In above-described embodiment of the application, perform step S110: according to step-out time analysis model and/or K-L quantity of information mould
Type carries out correlation calculations to warning index, to obtain the relative coefficient between each selected index and reference index, and
According to relative coefficient, selected index is screened, include walking as follows obtaining the process of leading indicators and coincidence indicator
Rapid:
The relative coefficient r between each selected index and reference index is obtained according to equation belowl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynOn the basis of)
Index, X=(x1,x2,…,xn) for being chosen index, l is the time difference, is in particular advanced or lag period, nlFor all indexs
Number;By time difference value in the first span and relative coefficient rlMore than first threshold selected index as in advance refer to
Mark, and by time difference value in the second span and relative coefficient rlMore than the selected index of Second Threshold as unanimously
Index.Wherein, the first span can be less than-3, and first threshold can be 0.7, and the second span may be greater than
In-2 and less than or equal to 2, Second Threshold can be 0.7.
Specifically, using reference index as " mark post " of screening in this step, model Preliminary screening is analyzed in used time difference correlation
Coincidence indicator and leading indicators.All selected the index the most advanced or delayed l phase in addition to reference index is (i.e. in above-described embodiment
The time difference) (l=0, ± 1, ± 2 ..., ± 12), calculate the dependency of each selected index and reference index the most respectively
Coefficient rl:
Y=(y in above formula1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) for being chosen index, r is phase relation
Number, l represents the advanced or lag period (i.e. the time difference), and l represents advanced when taking negative, represents delayed when taking positive number, l be referred to as the time difference or
Postpone number,WithIt is respectively sequence X and the meansigma methods of Y.nlIt it is the data amount check of all indexs.Then maximum time difference correlation coefficient
Being considered the time difference dependency relation reflecting selected index with reference index, the corresponding number l that postpones represents the advanced or lag period, i.e.
Make time difference relative coefficient rlMaximum delay number l is exactly this selected index and the advanced of reference index or lag period.
Fig. 5 is the method schematic diagram of the acquisition coincidence indicator according to embodiment illustrated in fig. 4 and leading indicators, specifically,
Get relative coefficient rlAfterwards, by the method acquisition leading indicators shown in Fig. 5 and coincidence indicator:
Step S302: the Testing index data whether time difference<-3 and relative coefficient>0.7 or achievement data whether-2≤
The time difference≤2 and relative coefficient > 0.7.Wherein, at the time difference<-3 and relative coefficient>0.7 or the achievement data of achievement data
Eligible-2≤time difference≤2 and relative coefficient > in the case of 0.7, perform step S304, do not meet the time difference at achievement data
<-3 and relative coefficient>0.7, and do not meet-2≤time difference≤2 and relative coefficient>in the case of 0.7, perform step S310:
Abandon achievement data.
Step S304: obtain initial leading indicators and initial coincidence indicator, wherein, in the time difference <-3 and the phase of achievement data
Close property coefficient > it is initial leading indicators by this data decimation in the case of 0.7, in-2≤time difference≤2 of achievement data and relevant
Property coefficient > in the case of 0.7, choosing this achievement data is initial coincidence indicator.
In above-described embodiment of the application, according to step-out time analysis model and/or K-L information computation, warning index is entered
Row correlation calculations, with the relative coefficient between each selected index and reference index, and according to relative coefficient to quilt
Selective goal screens, and includes obtaining the step of leading indicators and coincidence indicator: obtain each selected according to equation below
Select the relative coefficient r between index and reference indexl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynOn the basis of)
Index, X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlNumber for all indexs;Time difference value is taken first
In the range of value and relative coefficient rlMore than the selected index of first threshold as the Raw performance of leading indicators, and by the time difference
Value is in the second span and relative coefficient rlMore than initial as coincidence indicator of the selected index of Second Threshold
Index;It is standardized processing to the Raw performance of reference index, the Raw performance of leading indicators and coincidence indicator, to obtain
Standard basis index series pt, standard be chosen sequence q of indext, wherein, standard is chosen index and includes standard leading indicators
And standard coincidence indicator;Obtain the K-L information that each standard is chosen between index and standard basis index as follows
Amount kl:
kl=∑ptln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12, T=1,
2 ..., n, l are the time difference, nlNumber for all indexs;By time difference value in the 3rd span and K-L quantity of information klIt is less than
The standard of the 3rd threshold value is chosen index as leading indicators, and by time difference value in the 4th span and K-L quantity of information
klLess than the selected index of the 4th threshold value as coincidence indicator.Wherein, the 3rd span can be less than-3, the 3rd threshold value
Can be 0.3, the 4th span may be greater than equal to-2 and less than or equal to 2, and the 4th threshold value can be 0.3.
Said method can be achieved by the steps of: reference index is done standardization, the standard basis after process
Index series is designated as pt:
Leading indicators and the coincidence indicator of primary election are also done standardization, and the standard after process is chosen the sequence of index
It is designated as qt:
Then, the K-L quantity of information k about reference index after each primary election beacon delay l it is calculated as followsl:
kl=∑ptln(pt/qt+l)=0, ± 1 ..., ± 12
Wherein, l represents the advanced or lag period, and l represents advanced when taking negative, represents delayed, when l is referred to as when taking positive number
Difference, nlBeing the data amount check (numbers of the most all indexs) after data are evened up, the t in above-mentioned formula represents month, and i represents week
Number.
After calculating 2L+1 K-L quantity of information, from klValue is selected minima kl′As selected index x about base
The K-L quantity of information of quasi-index y, i.e.Its corresponding delay number l*It is exactly that selected index is optimal advanced or stagnant
Rear moon number (season), wherein, K-L quantity of information is closer to 0, illustrate index x and reference index y closer to.
Specifically can screen leading indicators as follows and coincidence indicator is designated as W respectively(3)And Z(3):
Step S306: detect the initial leading indicators whether time difference <-3 and K-L quantity of information < and 0.3 or initial consistent data be
No-2≤time difference≤2 and K-L quantity of information < 0.3.Wherein, initial leading indicators meet the time difference <-3 and K-L quantity of information < 0.3 or
The initial consistent data of person meets-2≤time difference≤2 and K-L quantity of information < in the case of 0.3, performs step S308, initial leading
Index do not meet the time difference <-3 and K-L quantity of information < and 0.3, perform step S310: abandon achievement data, be not inconsistent at initial consistent data
Close-2≤time difference≤2 and K-L quantity of information < in the case of 0.3, performs step S310: abandon achievement data.
Step S308: selected leading indicators and coincidence indicator.Wherein, in the time difference <-the 3 and K-L information of initial leading indicators
< in the case of 0.3, this index selected is existing index to amount, at initial coincidence indicator data ,-2≤time difference≤2 and K-L quantity of information
< in the case of 0.3, this index selected is coincidence indicator.Such as: choosing delay number l=-2, the index of-3 ,-4 is as the elder generation of primary election
Row index, postpones number l=-1, and the index of 0,1 is as the coincidence indicator of primary election.
According to above-described embodiment of the application, according to composite index number model, leading indicators and coincidence indicator are synthesized,
Include using acquisition as the leading composite index number of early-warning parameters and the step of coincident composite Index: to leading indicators and coincidence indicator
Carry out symmetrical change process respectively, to obtain leading indicators symmetry rate of change CW, i(t) and coincidence indicator symmetry rate of change CZ, i
T (), wherein, carries out symmetrical change process by equation below to leading indicators, to obtain leading indicators symmetry rate of change Cw,i
(t):
Wherein,Be i-th (i=1,2 ..., kw) individual leading indicators, t=2,
3 ..., n, kwThe number of leading indicators;By equation below, coincidence indicator is carried out symmetrical change process, to obtain coincidence indicator
Symmetrical rate of change Cz,i(t):
Wherein,Be i-th (i=1,2 ..., kz) individual coincidence indicator, t=2,
3 ..., n, kzIt it is the number of coincidence indicator;To leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change Cz,i
T result that () is standardized processing and obtaining after trend adjustment carries out composite calulation, to obtain leading composite index number and
Cause composite index number.Wherein, the t in above-mentioned formula represents month.
Specifically, the symmetrical rate of change of index is sought and by its standardization with index synthetic model, according to following formula respectively
RightWithSeek symmetrical rate of change Cw,i(t) and Cz,i(t):
T=2,3 ..., n, wherein,Be i-th (i=1,2 ..., kw) individual elder generation
Row index,Be i-th (i=1,2 ..., kz) individual coincidence indicator, kwAnd kzIt is the number of leading indicators and coincidence indicator respectively.
In above-described embodiment of the application, to leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change
Cz,iT () is standardized processing and after trend adjustment, the result that obtains carries out composite calulation, with obtain leading composite index number and
The step of coincident composite Index includes: obtain normalization factor A by equation belowW, iAnd AZ, i: t=2,3,…,n;Use normalization factor AW, iAnd AZ, iRespectively by leading indicators symmetry rate of change Cw,i(t)
With coincidence indicator symmetry rate of change Cz,iT () is standardized processing, to obtain standardization rate of change Sw,i(t) and Sz,i(t), its
In,
To standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, to obtain the standardization of leading indicators
Average rate of change VwStandardization average rate of change V of (t) and coincidence indicatorz(t);Standardization mean change according to leading indicators
Rate VwStandardization average rate of change V of (t) and coincidence indicatorzT () carries out composite calulation, to obtain leading composite index number Iw(t) and
Coincident composite Index Iz(t), wherein, And Iw(1)=
100, Iz(1)=100。
Specifically, according to equation below to leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change Cz,i
T () does standardized calculation so that it is average absolute value is equal to 1, obtains normalization factor Aw,iAnd AZ, i:
It, T=2,3 ..., n,
Then according to Aw,iAnd AZ, iRespectively by Cw,i(t) and CZ, iT () standardization, obtains standardization rate of change Sw,i(t) and
Sz,i(t):
T=2,3 ..., n, afterwards according to standardization rate of change Sw,i(t) and Sz,i
(t) be averaged rate of change process, to obtain standardization average rate of change V of leading indicatorswThe standardization of (t) and coincidence indicator
Average rate of change Vz(t), and according to standardization average rate of change V of leading indicatorswT the standardization of () and coincidence indicator averagely becomes
Rate VzT () carries out composite calulation, to obtain leading composite index number Iw(t) and coincident composite Index Iz(t), specifically: Iw(1)=
100, Iz(1)=100, then
More specifically, getting leading composite index number Iw(t) and coincident composite Index IzAfter (t), electric power can be needed
Asking trend to be adjusted, method is as follows:
According to following compound interest formula each sequence of coincidence indicator group obtained respectively respective average rate of increase:
Wherein, as shown in Figure 6, t is month,WithIt is consistent respectively
Index group i-th index at first with the most metacyclic meansigma methods, mIiWith mLiBe respectively coincidence indicator group i-th index at first with
The most metacyclic moon number, k2It is coincidence indicator number, miIt is the center circulated at first to the moon number between the most metacyclic center.
Then the average rate of increase G of coincidence indicator group is obtainedr, and as target trend:Afterwards
To the leading and initial composite index number I of coincidence indicatorw(t) and IzT () obtains their respective balanced growth with compound interest formula respectively
Rate r 'wWith r 'z:
Wherein,
Standardization average rate of change V to leading indicators group and coincidence indicator group againw(t) and VzT () does trend adjustment:
V′w(t)=Vw(t)+(Gr-r′w), V 'z(t)=Vz(t)+(Gr-r′z), t=2,3 ..., n.Then according to above-mentioned enforcement
Method in example calculates composite index number: make I 'w(1)=100, I 'z(1)=100, then
Generate with leading composite index number CI that the benchmark time is 100w(t) and coincident composite Index CIz(t):
WhereinWithIt is I ' respectivelyw(t) and I 'zT () is in the meansigma methods in benchmark time.
According to above-described embodiment of the application, to standardization rate of change Sw,i(t) and Sz,iT () is averaged at rate of change
Reason, to obtain standardization average rate of change V of leading indicatorswStandardization average rate of change V of (t) and coincidence indicatorzThe step of (t)
Suddenly include:
By equation below respectively by the standardization rate of change S of leading indicatorsw,iThe standardization change of (t) and coincidence indicator
Rate Sz,i(t) be averaged rate of change process, to obtain average rate of change R of leading indicatorswThe average change of (t) and coincidence indicator
Rate Rz(t):
Specifically, initial composite index number I is being obtainedw(t) and IzBefore (t), by equation below respectively by leading indicators
Standardization rate of change Sw,iThe standardization rate of change S of (t) and coincidence indicatorz,iT the () rate of change that is averaged processes, to obtain in advance
Average rate of change R of indexwAverage rate of change R of (t) and coincidence indicatorz(t):
Then according to equation below gauge index normalization factor Fw:
Finally according to criterion factor FwIt is standardized the average rate of change to process, to obtain the standard of leading indicators
Change average rate of change VwStandardization average rate of change V of (t) and coincidence indicatorz(t):
Vw(t)=Rw(t)/Fw, Vz(t)=Rz(t), t=2,3 ..., n, wherein, by the average rate of change of coincidence indicator sequence
Amplitude goes to adjust leading indicators sequence and the average rate of change of lagging indicator sequence, its purpose is to two indexes as one
Individual harmonious system is applied.
It should be noted that can be at such as one group of computer executable instructions in the step shown in the flow chart of accompanying drawing
Computer system performs, and, although show logical order in flow charts, but in some cases, can be with not
It is same as the step shown or described by order execution herein.
As can be seen from the above description, present invention achieves following technique effect: by the acquisition electric power of the application
The method and device of the early-warning parameters of demand, after the trend term obtained in original data sequence and periodic term, passes through logarithm
Obtain leading indicators and coincidence indicator according to sequence screening and calculating, then above-mentioned leading indicators and coincidence indicator index are synthesized
Model synthesis obtain early-warning parameters, and according to early-warning parameters analyze power demand cycle fluctuation, solve in prior art
Electricity needs field uses Predicting Technique to obtain short-term cyclic fluctuation distortion, it is impossible to acquire the early warning meeting electricity needs
Parameter, thus cause formulating the measure reasonably tackling cyclic fluctuation according to power cycle fluctuation, it is achieved that accurately
Obtain the early-warning parameters of electricity needs, thus formulate the effect of the counter-measure of reasonable science accurately according to short-term cyclic fluctuation
Really, and then slow down the amplitude of cyclic swing, reduce the destructiveness that power industry and economic development are caused by cyclic swing.
Obviously, those skilled in the art should be understood that each module of the above-mentioned present invention or each step can be with general
Calculating device realize, they can concentrate on single calculating device, or be distributed in multiple calculating device and formed
Network on, alternatively, they can with calculate the executable program code of device realize, it is thus possible to by they store
Performed by calculating device in the storage device, or they are fabricated to respectively each integrated circuit modules, or by them
In multiple modules or step be fabricated to single integrated circuit module and realize.So, the present invention be not restricted to any specifically
Hardware and software combines.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for the skill of this area
For art personnel, the present invention can have various modifications and variations.All within the spirit and principles in the present invention, that is made any repaiies
Change, equivalent, improvement etc., should be included within the scope of the present invention.
Claims (12)
1. the method for the early-warning parameters obtaining electricity needs, it is characterised in that including:
Obtain the data sequence for generating warning index;
According to adjusting parameter, described data sequence is screened, to obtain the data sequence including trend term and periodic term;
The Trend index of the data sequence of described trend term and periodic term is included described in calculating, and according to described Trend index pair
The described data sequence including described trend term and periodic term filters, and to obtain warning index sequence, described early warning refers to
Mark sequence is that in described data sequence, Trend index is the data increased;
Extract index on the basis of the generated energy in described warning index sequence, and the index that extraction is in addition to described generated energy is quilt
Selective goal;
According to step-out time analysis model and/or K-L information computation, described warning index is carried out correlation calculations, each to obtain
Relative coefficient between described selected index and described reference index, and be chosen described according to described relative coefficient
Index is screened, to obtain leading indicators and coincidence indicator;
According to composite index number model, described leading indicators and coincidence indicator are synthesized, to obtain leading as early-warning parameters
Composite index number and coincident composite Index;
After obtaining the data sequence for generating warning index, described method also includes: to the number in described data sequence
According to carrying out pretreatment, described pretreatment includes: fill up missing data process, revise noise data process, data smoothing process with
And data normalization processes.
Method the most according to claim 1, it is characterised in that utilize step-out time analysis model and/or K-L information computation pair
Described warning index carries out correlation calculations, with the dependency system between each described selected index and described reference index
Number, and according to described relative coefficient, described selected index is screened, to obtain the step of leading indicators and coincidence indicator
Suddenly include:
The relative coefficient r between each described selected index and described reference index is obtained according to equation belowl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) it is selected
Select index,WithBeing respectively sequence X and the meansigma methods of Y, l is the time difference, nlFor the number of all indexs, t=1,2 ..., n is the moon
Number, xt+1For the selected index of the t+1 month, ytReference index for the t month;
By described time difference value in the first span and described relative coefficient rlMake more than the selected index of first threshold
For described leading indicators, and by described time difference value in the second span and described relative coefficient rlMore than Second Threshold
Selected index as described coincidence indicator.
Method the most according to claim 2, it is characterised in that according to step-out time analysis model and/or K-L information computation pair
Described warning index carries out correlation calculations, with the dependency system between each described selected index and described reference index
Number, and according to described relative coefficient, described selected index is screened, to obtain the step of leading indicators and coincidence indicator
Suddenly include:
The relative coefficient r between each described selected index and described reference index is obtained according to equation belowl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynIndex on the basis of), X=(x1,x2,…,xn) it is selected
Selecting index, l is the time difference, nlFor the number of all indexs, t=1,2 ..., n is moon number;
By described time difference value in the first span and described relative coefficient rlMake more than the selected index of first threshold
For the Raw performance of described leading indicators, and by described time difference value in the second span and described relative coefficient rlGreatly
In the selected index of Second Threshold as the Raw performance of described coincidence indicator;
The Raw performance of described reference index, the Raw performance of described leading indicators and described coincidence indicator is standardized
Process, to obtain standard basis index series pt, standard be chosen sequence q of indext, wherein, described standard is chosen index bag
Include standard leading indicators and standard coincidence indicator;
Obtain the K-L quantity of information k that each standard is chosen between index and described standard basis index as followsl:
kl=∑ ptln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12,T=1,
2 ..., n is moon number, and l is the time difference, nlNumber for all indexs;
By described time difference value in the 3rd span and described K-L quantity of information klIt is chosen to refer to less than the standard of the 3rd threshold value
It is denoted as described leading indicators, and by described time difference value in the 4th span and described K-L quantity of information klLess than the 4th
The selected index of threshold value is as described coincidence indicator.
The most according to the method in claim 2 or 3, it is characterised in that according to composite index number model by described leading indicators and
Coincidence indicator synthesizes, and includes using acquisition as the leading composite index number of early-warning parameters and the step of coincident composite Index:
Described leading indicators and described coincidence indicator are carried out respectively symmetrical change process, to obtain leading indicators symmetry rate of change
Cw,i(t) and coincidence indicator symmetry rate of change Cz,i(t),
Wherein, by equation below, described leading indicators is carried out symmetrical change process, to obtain the change of described leading indicators symmetry
Rate Cw,i(t):
Wherein,Be i-th (i=1,2 ..., kw) individual leading indicators, t=2,
3 ..., n, kwNumber for leading indicators;
By equation below, described coincidence indicator is carried out symmetrical change process, to obtain coincidence indicator symmetry rate of change Cz,i
(t):
Wherein,Be i-th (i=1,2 ..., kz) individual coincidence indicator, t=2,
3 ..., n is moon number, kzIt it is the number of coincidence indicator;
To described leading indicators symmetry rate of change Cw,i(t) and described coincidence indicator symmetry rate of change Cz,iT () is standardized place
The result obtained after reason and trend adjustment carries out composite calulation, to obtain described leading composite index number and coincident composite Index.
Method the most according to claim 4, it is characterised in that to described leading indicators symmetry rate of change Cw,iT () is with described
Coincidence indicator symmetry rate of change Cz,iT result that () is standardized processing and obtaining after trend adjustment carries out composite calulation, with
The step obtaining described leading composite index number and coincident composite Index includes:
Normalization factor A is obtained by equation beloww,iAnd Az,i:T=2,
3,…,n;
Use described normalization factor Aw,iAnd Az,iRespectively by described leading indicators symmetry rate of change Cw,i(t) and described consistent finger
The symmetrical rate of change C of markz,iT () is standardized processing, to obtain standardization rate of change Sw,i(t) and Sz,i(t), wherein,
To described standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, to obtain the mark of described leading indicators
Standardization average rate of change VwStandardization average rate of change V of (t) and described coincidence indicatorz(t);
Standardization average rate of change V according to described leading indicatorswStandardization average rate of change V of (t) and described coincidence indicatorz
T () carries out composite calulation, to obtain described leading composite index number Iw(t) and coincident composite Index Iz(t), wherein,And Iw(1)=100, Iz(1)=100.
Method the most according to claim 5, it is characterised in that to described standardization rate of change Sw,i(t) and Sz,i(t) carry out
The average rate of change processes, to obtain standardization average rate of change V of described leading indicatorswThe standard of (t) and described coincidence indicator
Change average rate of change VzT the step of () including:
By equation below respectively by the standardization rate of change S of described leading indicatorsw,iThe standardization of (t) and described coincidence indicator
Rate of change Sz,i(t) be averaged rate of change process, to obtain average rate of change R of described leading indicatorswT () is consistent with described
Average rate of change R of indexz(t):
Wherein, λw,iAnd λz,iIt is leading indicators and coincidence indicator respectively
The weight of i-th index;
Criterion factor F is obtained by equation beloww:
According to described criterion factor FwIt is standardized the average rate of change to process, to obtain the standard of described leading indicators
Change average rate of change VwStandardization average rate of change V of (t) and described coincidence indicatorz(t), wherein, Vw(t)=Rw(t)/Fw, Vz
(t)=Rz(t)。
7. the device of the early-warning parameters obtaining electricity needs, it is characterised in that including:
First acquisition module, for obtaining the data sequence for generating warning index;
First processing module, for according to adjust parameter described data sequence is screened, with obtain include trend term and
The data sequence of periodic term;
First computing module, is used for including described in calculating the Trend index of the data sequence of described trend term and periodic term, and
According to described Trend index, the described data sequence including described trend term and periodic term is filtered, refer to obtaining early warning
Mark sequence, described warning index sequence is that in described data sequence, Trend index is the data increased;
First extraction module, for extracting index on the basis of the generated energy in described warning index sequence, and extracts except described
Index beyond electricity is for being chosen index;
Second computing module, for being correlated with to described warning index according to step-out time analysis model and/or K-L information computation
Property calculate, to obtain the relative coefficient between each described selected index and described reference index, and according to described relevant
Described selected index is screened by property coefficient, to obtain leading indicators and coincidence indicator;
Second processing module, for synthesizing described leading indicators and coincidence indicator according to composite index number model, to obtain
Leading composite index number and coincident composite Index as early-warning parameters;
Described device also includes: the tenth sub-processing module, for the data in described data sequence are carried out pretreatment, described pre-
Process includes: fills up missing data and processes, revises noise data process, data smoothing process and data normalization process.
Device the most according to claim 7, it is characterised in that described second computing module includes:
First sub-computing module, for obtaining between each described selected index and described reference index according to equation below
Relative coefficient rl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynRefer on the basis of)
Mark, X=(x1,x2,…,xn) for being chosen index,WithBeing respectively sequence X and the meansigma methods of Y, l is the time difference, nlFor all fingers
Target number, t=1,2 ..., n is moon number, xt+1For the selected index of the t+1 month, ytReference index for the t month;
First sub-processing module, for by described time difference value in the first span and described relative coefficient rlMore than
The selected index of one threshold value is as described leading indicators, and by described time difference value in the second span and described relevant
Property coefficient rlMore than the selected index of Second Threshold as described coincidence indicator.
Device the most according to claim 8, it is characterised in that described second computing module includes:
Second sub-computing module, for obtaining between each described selected index and described reference index according to equation below
Relative coefficient rl:
Wherein, l=0, ± 1, ± 2 ..., ± 12, Y=(y1,y2,…,ynRefer on the basis of)
Mark, X=(x1,x2,…,xn) for being chosen index, l is the time difference, nlFor the number of all indexs, t=1,2 ..., n is month
Number;
Second sub-processing module, for by described time difference value in the first span and described relative coefficient rlMore than
The selected index of one threshold value is as the Raw performance of described leading indicators, and by described time difference value in the second span
And described relative coefficient rlMore than the selected index of Second Threshold as the Raw performance of described coincidence indicator;
3rd sub-processing module, for described reference index, the Raw performance of described leading indicators and described coincidence indicator
Raw performance be standardized process, to obtain standard basis index series pt, standard be chosen sequence q of indext, wherein,
Described standard is chosen index and includes standard leading indicators and standard coincidence indicator;
3rd sub-computing module, is chosen between index and described standard basis index for obtaining each standard as follows
K-L quantity of information kl:
kl=∑ pt ln(pt/qt+1), wherein, l=0, ± 1 ..., ± 12, T=
1,2 ..., n is moon number, and l is the time difference, nlNumber for all indexs;
4th sub-processing module, for by described time difference value in the 3rd span and described K-L quantity of information klLess than the 3rd
The standard of threshold value is chosen index as described leading indicators, and by described time difference value in the 4th span and described K-
L quantity of information klLess than the selected index of the 4th threshold value as described coincidence indicator.
Device the most according to claim 8 or claim 9, it is characterised in that described second processing module includes:
5th sub-processing module, for carrying out symmetrical change process respectively, to obtain to described leading indicators and described coincidence indicator
Take leading indicators symmetry rate of change Cw,i(t) and coincidence indicator symmetry rate of change Cz,i(t), described 5th sub-processing module includes:
4th sub-computing module, for described leading indicators being carried out symmetrical change process by equation below, described to obtain
Leading indicators symmetry rate of change Cw,i(t):
Wherein,Be i-th (i=1,2 ..., kw) individual leading indicators, t=2,
3 ..., n is moon number, kwNumber for leading indicators;
5th sub-computing module, for described coincidence indicator being carried out symmetrical change process by equation below, consistent to obtain
Index symmetry rate of change Cz,i(t):
Wherein,Be i-th (i=1,2 ..., kz) individual coincidence indicator, t=2,
3 ..., n, kzIt it is the number of coincidence indicator;
6th sub-processing module, for described leading indicators symmetry rate of change Cw,i(t) and described coincidence indicator symmetry rate of change
Cz,iT () is standardized processing and after trend adjustment, the result that obtains carries out composite calulation, refers to obtaining described leading synthesis
Number and coincident composite Index.
11. devices according to claim 10, it is characterised in that described 6th sub-processing module includes:
6th sub-computing module, for obtaining normalization factor A by equation beloww,iAnd Az,i: T=2,3 ..., n;
7th sub-processing module, is used for using described normalization factor Aw,iAnd Az,iRespectively by described leading indicators symmetry rate of change
Cw,i(t) and described coincidence indicator symmetry rate of change Cz,iT () is standardized processing, to obtain standardization rate of change Sw,i(t) and
Sz,i(t), wherein,
8th sub-processing module, for described standardization rate of change Sw,i(t) and Sz,i(t) be averaged rate of change process, with
Obtain standardization average rate of change V of described leading indicatorswStandardization average rate of change V of (t) and described coincidence indicatorz(t);
7th sub-computing module, for standardization average rate of change V according to described leading indicatorsw(t) and described coincidence indicator
Standardization average rate of change VzT () carries out composite calulation, to obtain described leading composite index number Iw(t) and coincident composite Index Iz
(t), wherein,And Iw(1)=100, Iz(1)=
100。
12. devices according to claim 11, it is characterised in that described 8th sub-processing module includes:
9th sub-processing module, is used for by equation below respectively by the standardization rate of change S of described leading indicatorsw,i(t) and institute
State the standardization rate of change S of coincidence indicatorz,i(t) be averaged rate of change process, to obtain the mean change of described leading indicators
Rate RwAverage rate of change R of (t) and described coincidence indicatorz(t):
Wherein, λw,iAnd λz,iIt is leading indicators and coincidence indicator respectively
The weight of i-th index;
8th sub-computing module, for obtaining criterion factor F by equation beloww:
9th sub-computing module, for according to described criterion factor FwIt is standardized the average rate of change to process, to obtain
Standardization average rate of change V of described leading indicatorswStandardization average rate of change V of (t) and described coincidence indicatorz(t), wherein,
Vw(t)=Rw(t)/Fw, Vz(t)=Rz(t)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210425468.6A CN102930155B (en) | 2012-10-30 | 2012-10-30 | Obtain the method and device of the early-warning parameters of electricity needs |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210425468.6A CN102930155B (en) | 2012-10-30 | 2012-10-30 | Obtain the method and device of the early-warning parameters of electricity needs |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102930155A CN102930155A (en) | 2013-02-13 |
CN102930155B true CN102930155B (en) | 2016-10-05 |
Family
ID=47644952
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210425468.6A Active CN102930155B (en) | 2012-10-30 | 2012-10-30 | Obtain the method and device of the early-warning parameters of electricity needs |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102930155B (en) |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104182800B (en) * | 2013-05-21 | 2018-01-16 | 中国农业科学院棉花研究所 | The intelligent Forecasting of time series based on trend and cyclic swing |
CN104951991A (en) * | 2014-03-28 | 2015-09-30 | 中国石油天然气股份有限公司 | Method for acquiring user natural gas use behavior characteristics and device thereof |
CN103996076B (en) * | 2014-05-15 | 2017-11-21 | 广东电网公司 | The change method for early warning and system of electricity needs |
CN104965108A (en) * | 2015-05-25 | 2015-10-07 | 上海电机学院 | Intelligent ammeter used for early warning of step electric quantity violation and voltage violation, and early warning method |
CN110032745A (en) * | 2018-01-11 | 2019-07-19 | 富士通株式会社 | Generate the method and apparatus and computer readable storage medium of sensing data |
CN109754114A (en) * | 2018-11-30 | 2019-05-14 | 跨越速运集团有限公司 | Goods amount intelligent Forecasting and system |
CN111539844B (en) * | 2020-04-20 | 2023-06-16 | 海南电网有限责任公司 | Moon electricity quantity prediction method based on K-L information quantity and ARIMA error correction |
CN111476030B (en) * | 2020-05-08 | 2022-03-15 | 中国科学院计算机网络信息中心 | Prospective factor screening method based on deep learning |
CN112116139A (en) * | 2020-09-03 | 2020-12-22 | 国网经济技术研究院有限公司 | Power demand prediction method and system |
CN113344737B (en) * | 2021-06-04 | 2023-11-24 | 北京国电通网络技术有限公司 | Device control method, device, electronic device and computer readable medium |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1585953A (en) * | 2001-09-13 | 2005-02-23 | Abb股份有限公司 | Method and system to calculate a demand for electric power |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2008111467A1 (en) * | 2007-03-09 | 2008-09-18 | Sanyo Electric Co., Ltd. | Demand control system, demand controller, demand program, and demand control method |
-
2012
- 2012-10-30 CN CN201210425468.6A patent/CN102930155B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1585953A (en) * | 2001-09-13 | 2005-02-23 | Abb股份有限公司 | Method and system to calculate a demand for electric power |
Non-Patent Citations (2)
Title |
---|
基于景气指数的宏观经济监测预警***研究;龚盈盈;《中国优秀硕士学位论文全文数据库》;20060415(第4期);第12-18页 * |
短期电力供需预警***及应用;温权等;《能源技术经济》;20110120;第23卷(第1期);第25-30页 * |
Also Published As
Publication number | Publication date |
---|---|
CN102930155A (en) | 2013-02-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102930155B (en) | Obtain the method and device of the early-warning parameters of electricity needs | |
Yang et al. | A hybrid forecasting approach applied in wind speed forecasting based on a data processing strategy and an optimized artificial intelligence algorithm | |
Sriram et al. | Causal Markov Elman network for load forecasting in multinetwork systems | |
CN107622329A (en) | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales | |
CN105005825B (en) | A kind of method and system based on dynamic neural network prediction photovoltaic power | |
CN110212524A (en) | A kind of region Methods of electric load forecasting | |
CN111680841B (en) | Short-term load prediction method, system and terminal equipment based on principal component analysis | |
CN107563565A (en) | A kind of short-term photovoltaic for considering Meteorology Factor Change decomposes Forecasting Methodology | |
CN104699890B (en) | A kind of modeling method of the short-term Early-warning Model of city net distribution transforming heavy-overload | |
CN102495937A (en) | Prediction method based on time sequence | |
Toubeau et al. | Capturing spatio-temporal dependencies in the probabilistic forecasting of distribution locational marginal prices | |
CN113837499A (en) | Ultra-short-term wind power prediction method and system | |
CN106803135A (en) | The Forecasting Methodology and device of a kind of photovoltaic power generation system output power | |
CN103793887A (en) | Short-term electrical load on-line predicting method based on self-adaptation enhancing algorithm | |
Li et al. | A k-nearest neighbor locally weighted regression method for short-term traffic flow forecasting | |
CN104834975A (en) | Power network load factor prediction method based on intelligent algorithm optimization combination | |
CN105279582B (en) | Super short-period wind power prediction technique based on dynamic correlation feature | |
CN115238854A (en) | Short-term load prediction method based on TCN-LSTM-AM | |
Souhe et al. | Forecasting of electrical energy consumption of households in a smart grid | |
CN114595891A (en) | Power distribution network voltage and power flow boundary crossing risk assessment method, system and equipment | |
Usman et al. | Data analytics for short term price and load forecasting in smart grids using enhanced recurrent neural network | |
CN105260944A (en) | Method for calculating statistical line loss based on LSSVM (Least Square Support Vector Machine) algorithm and association rule mining | |
CN115860232A (en) | Steam load prediction method, system, electronic device and medium | |
CN112508734B (en) | Method and device for predicting power generation capacity of power enterprise based on convolutional neural network | |
CN115495862A (en) | Power transmission network extension planning method and system considering extreme scenes of renewable energy sources |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |