CN106408119A - Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas - Google Patents
Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas Download PDFInfo
- Publication number
- CN106408119A CN106408119A CN201610815916.1A CN201610815916A CN106408119A CN 106408119 A CN106408119 A CN 106408119A CN 201610815916 A CN201610815916 A CN 201610815916A CN 106408119 A CN106408119 A CN 106408119A
- Authority
- CN
- China
- Prior art keywords
- region
- moment point
- load
- weather
- moment
- 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.)
- Pending
Links
- 238000011156 evaluation Methods 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000005516 engineering process Methods 0.000 claims abstract description 10
- 238000006467 substitution reaction Methods 0.000 claims abstract description 10
- 230000001360 synchronised effect Effects 0.000 claims abstract description 10
- 238000011068 loading method Methods 0.000 claims description 62
- 239000011159 matrix material Substances 0.000 claims description 33
- 230000015572 biosynthetic process Effects 0.000 claims description 31
- 238000003786 synthesis reaction Methods 0.000 claims description 25
- 238000013277 forecasting method Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 11
- 238000009499 grossing Methods 0.000 claims description 9
- 241000083513 Punctum Species 0.000 claims description 3
- 230000008034 disappearance Effects 0.000 claims description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims description 2
- 238000012913 prioritisation Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 240000002853 Nelumbo nucifera Species 0.000 description 4
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 4
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000005611 electricity Effects 0.000 description 3
- 230000001186 cumulative effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000013439 planning Methods 0.000 description 2
- 238000006116 polymerization reaction Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas. The large power grid load prediction method includes the steps of obtaining a load prediction value and resent historical data of each subarea, dividing the whole network into N meteorological subareas by using synchronous back substitution eliminating technology, calculating average proportional coefficient of moment t in N subareas recently, predicting proportional coefficient of the moment t at the day to be predicted in the N subareas, constructing the comprehensive evaluation index of the moment t, selecting q subareas with higher priority, and predicting the whole network load of the moment t respectively, establishing an optimal comprehensive model for q different prediction values, obtaining the prediction value of the whole network load at the moment t, respectively establishing optimal comprehensive models for T moments in a day to obtain load prediction sequences, providing G load prediction sequences for G prediction schemes for different values q, and reestablishing an optimal comprehensive model for each moment to obtain final prediction result of the whole network load at the day to be predicted. Through the method, the short-period load prediction accuracy of a power system is improved.
Description
Technical field
The present invention relates to Load Prediction In Power Systems correlative technology field, specifically, it is related to a kind of class meteorology that is based on and divides
The bulk power grid load forecasting method of area's overall merit, for power-system short-term load forecasting.
Background technology
For ensureing the dynamic equilibrium of power system generated output and load power it is necessary to make science to power system load
Prediction.Load prediction is the important process that control centre and power network development plan department, and the result of load prediction is to electricity
The aspects such as network operation, control, scheduling, planning, construction have important guiding value, and it is electrical network scientific development and scientific dispatch
Basis.
Improve load prediction technical merit, be conducive to planned supply and use of electric power manage, be conducive to reasonable arrangement power system operating mode and
Unit maintenance scheduling, is conducive to economizing on coal, fuel-economizing and reduce cost of electricity-generating, is conducive to improving economic benefit and the society of power system
Benefit.Therefore, load prediction has become the important content realizing power system management modernization.
Currently with region load prediction the whole network system loading method be " subnet summation ", " subnet summation " pre-
Flow gauge is as shown in Figure 1.Its prediction is segmented into following three steps substantially:
1st, take into full account the factor of various impact short-term load forecastings in regional, select suitable Forecasting Methodology,
Then short-term load forecasting is carried out to regional according to each regional historical data;Draw 96 point load prediction knots of each subnet
Really.
2nd, each region load prediction results are collected, the 96 point prediction data in each region are carried out adding up obtains each moment
Point cumulative and.
3rd, 96 points of the station service of day to be predicted and network loss are calculated, and to cumulative and be modified drawing final Anhui electricity
Net load prediction results.
Because the method will be predicted to each region subnet, and for short-term load forecasting, due to each area
The steady load degree in domain is different, and prediction difficulty is totally different, and meanwhile, station service and network loss data are also required to predict, therefore, work as profit
When carrying out subnet with the prediction load of Zone Full and adding up, the accuracy effect of the whole network system loading prediction may be not ideal enough.
In order to solve the above problems, the applicant in the application for a patent for invention of Application No. CN201310648023.9,
Propose a kind of the whole network load forecasting method based on region predicted load overall merit accuracy rate, the method is to a certain degree
On improve the accuracy of power-system short-term load forecasting.But said method is when carrying out region division, it is with geographical position
Residing administrative region divides to region, but the variation tendency of each administrative region meteorological condition is different, especially
In meteorological condition change violent summer it is thus possible to lead to the load of each administrative region and the ratio of the load of the whole network connecting
Change greatly in continuous several days, be unfavorable for the prediction of each administrative region and the whole network load proportion coefficient;According to comprehensive evaluation index
Result select q regional prediction the whole network load, due to different q values, will have different prediction effects, therefore arrange different pre-
Survey scheme and its number of regions of selection, can avoid the presence of prediction limits error.
Content of the invention
The present invention be directed to existing when bulk power grid load being predicted based on region predicted load overall merit, adopt
Administrative region residing for geographical position carries out the problems of region division to bulk power grid, provides one kind based on class meteorology subregion
The bulk power grid load forecasting method of overall merit, to improve power-system short-term load forecasting accuracy rate.Meanwhile, this method is same
Selected part region predicted load, to predict the whole network load, can avoid some region load predictions specific responsibilitys to report regional prediction
Result affects the whole network load prediction not in time.
The technical problem solving required for the present invention, can be achieved through the following technical solutions:
A kind of bulk power grid load forecasting method based on class meteorology subregion overall merit is it is characterised in that include following walking
Suddenly:
(1) obtain the predicted load of the M administrative region having reported in daily load prediction value to be predicted;Obtain recent one
As historical data sample space, the load data in described historical data is the whole network and M to historical data in the individual sample period
The actual load of individual administrative region and prediction load;
(2) at least 2 history meteorological condition data in M administrative region in the sample period are obtained, meteorological by history
Condition data, is divided into the weather region that N number of meteorological condition differs greatly;
(3) calculate the average proportions coefficient in moment point t for the described N number of weather regionUse exponential smoothing dynamic prediction
Day to be predicted, N number of weather region, in the proportionality coefficient of identical moment point t, obtained the proportionality coefficient in moment point t for N number of weather region
Matrix Ct;
(4) by the weight of multiple single evaluation indexs and each single evaluation index, build moment point t overall merit and refer to
Mark FAL,t;
(5) N number of weather region is pressed to the comprehensive evaluation index F in each region in moment point tAL,tPriority from small to large
Sequence, and with comprehensive evaluation index FAL,tMinimum highest priority.
(6) (q is between 1 to arrange the weather region number q selecting in G (1≤G≤N) individual prediction scheme and each prediction scheme
Numerical value to N), and the numerical value of the q of each prediction scheme is different;
(7) for certain prediction scheme g, select q higher weather region of moment point t priority, with selected q
The whole network system loading of weather region difference prediction time point t, obtains q different predicting the outcome;
(8) q different predicting the outcome is set up with the optimal synthesis model of moment point t, calculates selected each meteorologic district
The optimal weights in domain, obtain moment point t the whole network system loading finally predict the outcome for
Wherein,It is the whole network system loading of being predicted by the weather region k of prediction scheme g optimum in moment point t
Weight,The whole network system loading of moment point t predicting out with k-th weather region for prediction scheme g;
(9) for day to be predicted whole day T moment point, set up optimal synthesis model respectively, obtain the whole day of prediction scheme g
Load prediction sequence
(10) for different the predicting the outcome of G prediction scheme G, the optimal synthesis of whole day T moment point are set up respectively
Model, obtains the final load prediction results (L of day to be predicted1,L2,…,LT):
Wherein,The whole network system loading predicting for prediction scheme g is in the optimal weights of moment point t.
In the present invention, described at least 2 history meteorological condition data are the highest temperature of administrative region in the sample period
2 in degree, minimum temperature, mean temperature, human comfort, weather pattern, wind speed, comfort index, coldness index or 2
Above any combination.
Described at least 2 history meteorological condition data are the maximum temperature of administrative region, lowest temperature in the sample period
Degree and mean temperature, the history meteorological condition data according to M administrative region forms meteorological data matrix X:
X=(X1,X2,…Xi,…,XM)
Wherein,
XiFor i-th administrative region within a sample period meteorological condition data acquisition system of n days,
For i-th administrative region within a sample period maximum temperature of n days;
For i-th administrative region within a sample period minimum temperature of n days;
For i-th administrative region within a sample period mean temperature of n days.
In the present invention, using the synchronous back substitution technology for eliminating based on probability metrics, matrix X is polymerized, is divided into N number of
The weather region that meteorological condition differs greatly:
First, a Probability p occurring is specified to M column vector in matrix Xs=1/M (s=1,2 ..., M).Set S table
Show initial administrative region, make ξsS-th administrative region in representing matrix X, DTs,s'Represent administrative region s and administrative region s'
Distance, its value is vector 2 norm between administrative region s and administrative region s', the vectorial DA=[1,2 ..., M] of settingT, DB
It is all the square formation of 0 M × M for all elements.
The basic step that synchronous back substitution eliminates is as follows:
1) calculate the distance between each pair administrative region DTs,s', DTs,s'=| | ξs,ξs'||.
2) for each administrative region k, find out the administrative region r, i.e. DT the shortest with administrative region k distancek(r)=
minDTk,s'.
3) calculate PDk(r)=pk*DTk(r), k ∈ S, finds out administrative region index h so that PD in kh=minPDk,k∈
S.
4) make S=S- ξDA(h), and pr=pr+ph.DB (h, h)=DA (r), DA (h) are set to sky.
5) 2 are repeated) -4), till remaining administrative region number is N.
DA is only left N number of element, corresponding square formation DB (DA (i), DA (i))=0, i=1,2 ... N, that is, need to retain N number of gas
As the weather region that condition difference is larger.
The administrative region search step of each auto polymerization in weather region that in DA, N number of meteorological condition differs greatly is as follows:
1) i=1, P=DA (i) are made.
2) find out and meet the m requiring so that DB (m, m)=P, then the administrative region comprising in the i of weather region have P=[P,
m].
3) find out and meet the l requiring so that DB (l, l)=m, P=[P, l], this search procedure is up to the diagonal element of DB
Till not being equal to l;
4) i=i+1, P=DA (i), repeat 2) -4), until i=N.
In the present invention, in described step (1), described historical data is pre-processed as follows:
Order:L (d, t) is the load value of the d days t, L (d, t1) and L (d, t2) it is the d days two adjacent with t
Moment t1、t2Load value, L (d1, t) with L (d2, t) it is the load value in moment point t for two adjacent with d day;
A) for the process of missing data
If load value L (d, the t) disappearance of the d days t, then formula (1) is utilized to obtain L (d, t):
L (d, t)=α L (d, t1)+αL(d,t2)+βL(d1,t)+βL(d2,t) (1)
In formula (1), α and β is coefficient, α>β, alpha+beta=1;
B) for the process of bad point data
Define the bias ratio that allows for load of ε, ρ (d, t) is the actual bias ratio of the d days moment point t, as ρ (d, t) >=ε,
Judge that L (d, t) is bad data, bad data is usedReplaced:
In the present invention, described step (3) is expressed as:
Ct=(C1,t,C2,t,…,CN,t)
Wherein,
For the 1st administrative region historical data moment point t average proportions coefficient;
For the 2nd administrative region historical data moment point t average proportions coefficient;
For n-th administrative region historical data moment point t average proportions coefficient;
C1,tFor the 1st administrative region day to be predicted moment point t proportionality coefficient;
C2,tFor the 2nd administrative region day to be predicted moment point t proportionality coefficient;
CN,tFor n-th administrative region day to be predicted moment point t proportionality coefficient.
Exponential smoothing in described step (3) is that onset index smoothing model is:
In formula (3):Represent that weather region i accounts for the predicted value of the whole network system loading ratio in moment t,J before expression
It weather region i accounts for the actual value of the whole network system loading ratio in moment t;N is the number of days within a sample period;λjTable
Show weight coefficient, λj=λ (1- λ)j-1, λ is constant, and 0<λ<1.
In the present invention, in described step (4), build multiple single evaluation index systems F of moment point tt, using three lists
One evaluation index, then the multiple criteria system F of each weather regiontFor:
Ft=(F1,t,F2,t,F3,t)T
Described three single evaluation indexs are respectively:Certain weather region load refers in the comprehensive stability degree evaluation of moment point t
Mark F1,t, certain weather region load is in prediction synthesis accuracy rate evaluation index F of moment point t2,t, certain weather region load proportion system
Number is in stability evaluation index F of moment point t3,t;
Build moment point t comprehensive evaluation index FAL,tFor;
FAL,t=ω × Y
Wherein:ω=(ω1,ω2,ω3), ω is the weight matrix in moment point t for three single evaluation indexs;ω1,ω2,
ω3For multiple single evaluation index systems FtIn each single evaluation index moment point t weight coefficient;Y is according to multiple lists
One assessment indicator system FtThe decision matrix in moment point t being formed.
Certain weather region load described is in comprehensive stability degree evaluation index F of moment point t1,tObtained by formula (4):
In formula (4):RSDtFor certain weather region moment t weather region load relative standard deviation,For certain
Weather region accounts for the average proportions coefficient of system loading in moment t time domain load;
Certain weather region load described is in prediction synthesis accuracy rate evaluation index F of moment point t2,tObtain by formula (5)
?:
In formula (5):QtConsensus forecast accuracy rate for moment t in selected sample space for certain weather region;
Certain weather region load proportion coefficient described is in stability evaluation index F of moment point t3,tObtain by formula (6):
In formula (6):For the sample canonical variance of the proportionality coefficient in moment t for certain weather region,For certain gas
Sample mathematic expectaion as the proportionality coefficient in moment t for the region;
Or, each single evaluation index described is in the weight coefficient ω of moment point t1,ω2,ω3And decision matrix Y presses
Following method obtains:
1) the decision matrix Y is made to be:Y=(yil)3×N, wherein:
In formula (7):Fi,lL-th evaluation index for weather region i, mini{Fi,lIt is in N number of weather region l-th
The minimum of a value of evaluation index, maxi{Fi,lBe l-th evaluation index in N number of weather region maximum;
2) then have:
In formula (8):Wherein slFor the standard variance of l item index in decision matrix Y,For l item index
Mathematic expectaion, ωlIt is l-th evaluation index weight coefficient in moment point t.
In the present invention, in described step (7), q different predicting the outcome is obtained for certain prediction scheme g, is expressed as:
Wherein,
It is the whole network system loading of the 1st moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of the 2nd moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of moment point t being predicted using q-th weather region of prediction scheme g, andLq,tFor q-th weather region moment point tt predicted load, Cq,tFor q-th weather region when
The proportionality coefficient predicted value of punctum t.
In the present invention, in described step (8), optimal synthesis model is set up as follows:
Characterize the object function of moment point t the whole network system loading predicted value with formula (9), formula (10)-(11) are object function
Constraints:
In formula (9):Represent that the whole network system that jth sky moment point t of prediction scheme g is predicted by weather region k is born
Lotus predicted value, Lall,t,jRepresent the whole network system loading actual value of jth sky moment point t;
Q weather region selected in prediction scheme g is calculated in moment point t by formula (9), formula (10) and formula (11)
Optimal weightsAfterwards, further according toThe whole network system that weighting obtains moment point t of prediction scheme g is born
Lotus predicted value.
In the present invention, in described step (4), the optimal synthesis model of each prediction scheme is set up as follows:
Characterize the object function of moment point t the whole network system loading predicted value with formula (12), formula (13)-(14) are object function
Constraints:
In formula (12):Represent the whole network system loading predicted value that jth sky moment point t is predicted by prediction scheme g,
Lall,t,jRepresent the whole network system loading actual value of jth sky moment point t.
The optimal weights in moment point t for the G weather region are calculated by formula (12), formula (13) and formula (14)Afterwards, then
According toWeighting obtains the final predicted value of the whole network system loading of day to be predicted moment point t.
, there is following beneficial effect in the bulk power grid load forecasting method based on class meteorology subregion overall merit for the present invention:
1st, to reduce load fluctuation in subnet summation in prior art larger and be difficult to the area predicting for the inventive method
The impact that domain is caused to the whole network load prediction;Avoid the prediction to power plant's electricity consumption and grid loss.
2nd, need in subnet summation to know all regions predicted load in advance, and the inventive method selected part area
Domain predicted load is predicting the whole network load, it is to avoid some region load predictions specific responsibilitys report regional prediction result not in time and
The whole network load prediction at impact province regulation and control center.
3rd, the inventive method adopts the method that overall target is evaluated that each area priorities are sorted, and can consider each
The impact to the whole network predicted load for the region predicted load, is conducive to improving the accuracy rate of the whole network load prediction.
4th, the division in region is no longer simply carried out using administrative region, by being based on meteorological condition similarity to administrative region
Processed, it is to avoid when each administrative region meteorological condition changes greatly, especially in the summer that meteorological condition change is violent
The ratio of season, the load of each administrative region that may lead to and the load of the whole network changed greatly the shadow of generation in continuously several days
Ring, by the division to weather region, be conducive to the prediction of each weather region and the whole network load proportion coefficient;Commented according to comprehensive
The result of valency index selects q regional prediction the whole network load, due to different q values, will have different prediction effects, therefore setting is not
Same prediction scheme and its number of regions of selection, can avoid the presence of prediction limits error.
Brief description
To further illustrate the present invention below in conjunction with the drawings and specific embodiments.
Fig. 1 is subnet accumulation algorithm flow process.
Fig. 2 is the inventive method flow chart.
Fig. 3 is that 6 prediction scheme that certain saves electrical network on May 18th, 2016 predict the 6 the whole network load predictions obtaining respectively
Curve.
Fig. 4 is load prediction curve and the realized load curve comparison diagram of certain province's electrical network on May 18th, 2016 final whole day.
Specific embodiment
In order that the technological means of the present invention, creation characteristic, reached purpose and effect are easy to understand, with reference to tool
Body illustrates, and the present invention is expanded on further.
Idea of the invention is that, by basic operation being predicted to bulk power grid load to based on region predicted load
The problems of the analysis of mode situation, find region to be carried out divide using the administrative region residing for geographical position, provide
A kind of bulk power grid load forecasting method based on class meteorology subregion overall merit is accurate to improve power-system short-term load forecasting
Rate.
Referring to Fig. 2, based on the bulk power grid load forecasting method of class meteorology subregion overall merit, comprise the following steps:
The historical data obtaining (such as n days) a recent sample period Nei is as historical data sample space, history number
According in load data be the actual load of M administrative region and predict load in the whole network and the whole network, here, administrative region refers to
The region being divided with geographical position, for example, in the case that the scope of the whole network is a province, administrative region can be this inside the province
The administrative division being carried out in units of city.
For described historical data, can be pre-processed as follows:
Order:L (d, t) is the load value of the d days t, L (d, t1) and L (d, t2) it is the d days two adjacent with t
Moment t1、t2Load value, L (d1, t) with L (d2, t) it is the load value in moment point t for two adjacent with d day;
A) for the process of missing data
If load value L (d, the t) disappearance of the d days t, then formula (1) is utilized to obtain L (d, t):
L (d, t)=α L (d, t1)+αL(d,t2)+βL(d1,t)+βL(d2,t) (1)
In formula (1), α and β is coefficient, α>β, alpha+beta=1;
B) for the process of bad point data
Define the bias ratio that allows for load of ε, ρ (d, t) is the actual bias ratio of the d days moment point t, as ρ (d, t) >=ε,
Judge that L (d, t) is bad data, bad data is usedReplaced:
Then at least 2 history meteorological condition data in M administrative region in the sample period, history meteorological condition are obtained
Data preferably adopts these three history of the maximum temperature of administrative region, minimum temperature and mean temperature in the sample period meteorological
Condition data, certainly, for lifting prediction accuracy demand, history meteorological condition data can also using human comfort,
Weather pattern, wind speed, comfort index, coldness index etc., are not the history meteorological condition data of data for weather pattern etc., can
So that data is converted into using data processing.For the ease of example, in present embodiment, adopt the maximum temperature of administrative region, minimum
Temperature and these three history meteorological condition data of mean temperature are illustrative.
History meteorological condition data according to M administrative region forms meteorological data matrix X:
X=(X1,X2,…Xi,…,XM)
Wherein,
XiFor i-th administrative region within a sample period meteorological condition data acquisition system of n days,
For i-th administrative region within a sample period maximum temperature of n days;
For i-th administrative region within a sample period minimum temperature of n days;
For i-th administrative region within a sample period mean temperature of n days.
So, the row vector number of X matrix is 3*n, and column vector number is M.
In the present invention, using the synchronous back substitution technology for eliminating based on probability metrics, matrix X is polymerized, divides N number of gas
As the weather region that condition difference is larger:
First, a Probability p occurring is specified to M column vector in matrix Xs=1/M (s=1,2 ..., M).Set S table
Show initial administrative region, make ξsS-th administrative region in representing matrix X, DTs,s' represent administrative region s and administrative region
The distance of s', its value is vector 2 norm between administrative region s and administrative region s', the vectorial DA=[1,2 ..., M] of settingT,
DB is the square formation of 0 M × M for all elements.
The basic step that synchronous back substitution eliminates is as follows:
1) calculate the distance between each pair administrative region DTs,s', DTs,s'=| | ξs,ξs'||.
2) for each administrative region k, find out the administrative region r, i.e. DT the shortest with administrative region k distancek(r)=
minDTk,s'.
3) calculate PDk(r)=pk*DTk(r), k ∈ S, finds out administrative region index h so that PD in kh=minPDk,k∈
S.
4) make S=S- ξDA(h), and pr=pr+ph.DB (h, h)=DA (r), DA (h) are set to sky.
5) 2 are repeated) -4), till remaining administrative region number is N.
DA is only left N number of element, corresponding square formation DB (DA (i), DA (i))=0, i=1,2 ... N, that is, need to retain N number of gas
As the weather region that condition difference is larger.
The administrative region search step of each auto polymerization in weather region that in DA, N number of meteorological condition differs greatly is as follows:
1) i=1, P=DA (i) are made.
2) find out and meet the m requiring so that DB (m, m)=P, then the administrative region comprising in the i of weather region have P=[P,
m].
3) find out and meet the l requiring so that DB (l, l)=m, P=[P, l], this search procedure is up to the diagonal element of DB
Till not being equal to l;
4) i=i+1, P=DA (i), repeat 2) -4), until i=N.
In fact, the essence being divided into the weather region that N number of meteorological condition differs greatly is, by gas in M administrative region
It is polymerized during prediction as the less region of condition difference and (will be considered as together the less administrative region of meteorological condition difference
One weather region)).Specifically when being operated, can be M administrative region is divided into based on the difference of meteorological condition N number of
Weather region, comprises one or several administrative regions in each weather region.
After completing the acquisition of weather region, calculate the average proportions coefficient in moment point t for N number of weather regionUse index
Exponential smoothing dynamic prediction day to be predicted N number of weather region identical moment point t proportionality coefficient, obtain N number of weather region when
The proportionality coefficient Matrix C of punctum tt;It is expressed as:
Ct=(C1,t,C2,t,…,CN,t)
Wherein,
For the 1st administrative region historical data moment point t average proportions coefficient;
For the 2nd administrative region historical data moment point t average proportions coefficient;
For n-th administrative region historical data moment point t average proportions coefficient;
C1,tFor the 1st administrative region day to be predicted moment point t proportionality coefficient;
C2,tFor the 2nd administrative region day to be predicted moment point t proportionality coefficient;
CN,tFor n-th administrative region day to be predicted moment point t proportionality coefficient.
Exponential smoothing is that onset index smoothing model is:
In formula (3):Represent that weather region i accounts for the predicted value of the whole network system loading ratio in moment t,J before expression
It weather region i accounts for the actual value of the whole network system loading ratio in moment t;N is the number of days within a sample period;λjTable
Show weight coefficient, λj=λ (1- λ)j-1, λ is constant, and 0<λ<1, the weight for ensureing Recent data is big, and weight at a specified future date is little, λ
Generally take the constant between 0.7~0.9.
By the weight of multiple single evaluation indexs and each single evaluation index, build moment point t comprehensive evaluation index
FAL,t, concrete operations are:Build multiple single evaluation index systems F of moment point tt, using three single evaluation indexs, then often
The multiple criteria system F of individual weather regiontFor:
Ft=(F1,t,F2,t,F3,t)T
Described three single evaluation indexs are respectively:Certain weather region load refers in the comprehensive stability degree evaluation of moment point t
Mark F1,t, certain weather region load is in prediction synthesis accuracy rate evaluation index F of moment point t2,t, certain weather region load proportion system
Number is in stability evaluation index F of moment point t3,t;
Due to containing three kinds of single evaluation indexs, three evaluation indexes in selected sample space in the multiple criteria system
It is possible that different evaluation results, therefore how these three indexs of overall merit, therefrom choose most suitable several meteorology
Region then becomes crucial predicting the whole network load, builds moment point t comprehensive evaluation index FAL,tFor;
FAL,t=ω × Y
Wherein:ω=(ω1,ω2,ω3), ω is the weight matrix in moment point t for three single evaluation indexs;ω1,ω2,
ω3For multiple single evaluation index systems FtIn each single evaluation index moment point t weight coefficient;Y is according to multiple lists
One assessment indicator system FtThe decision matrix in moment point t being formed.
Certain weather region load is in comprehensive stability degree evaluation index F of moment point t1,tObtained by formula (4):
In formula (4):RSDtFor certain weather region moment t weather region load relative standard deviation,For certain
Weather region weather region load in moment t accounts for the average proportions coefficient of system loading;Load comprehensive stability degree index F1,t's
Physical significance is:When with certain weather region load prediction the whole network load, the size of weather region load fluctuation is born in the whole network
The embodiment of lotus.When with load comprehensive stability degree index F1,tEach moment point is ranked up to each weather region, numerical value is
Little region, when representing in this moment point with load prediction the whole network load of this weather region, deviation is minimum.
Certain weather region load is in prediction synthesis accuracy rate evaluation index F of moment point t2,tObtain by formula (5):
In formula (5):QtConsensus forecast accuracy rate for moment t in selected sample space for certain weather region;Load
Comprehensive stability degree index F2,tPhysical significance be:When with certain weather region load prediction the whole network load, weather region load
The size of prediction deviation is in the embodiment of the whole network load.When with load comprehensive stability degree index F2,tTo each in each moment point
Weather region is ranked up, the minimum region of numerical value, and predictablity rate highest represents and is taken in this meteorologic district in this moment point
The predicted load in domain predicts the whole network load, and the predicated error of the whole network load is minimum.
When with weather region load prediction the whole network load, the whole network load not only quality with weather region predicted load
Relevant, and to account for the whole network load proportion coefficient predictors relevant with the weather region load in day to be predicted, therefore proposes gas here
As region load accounts for the whole network load proportion coefficient stabilization degree index.Certain weather region load proportion coefficient stablizing in moment point t
Degree evaluation index F3,tObtain by formula (6):
In formula (6):For the sample canonical variance of the proportionality coefficient in moment t for certain weather region,For certain gas
Sample mathematic expectaion as the proportionality coefficient in moment t for the region.
Each single evaluation index is in the weight coefficient ω of moment point t1,ω2,ω3Obtain as follows:
Because the dimension of three single evaluation indexs is different with the order of magnitude, carry out each single evaluation indices non-dimension first
Process obtains decision matrix Y:Y=(yil)3×N, wherein:
In formula (7):Fi,lL-th evaluation index for weather region i, mini{Fi,lIt is in N number of weather region l-th
The minimum of a value of evaluation index, maxi{Fi,lBe l-th evaluation index in N number of weather region maximum.
Then have:
In formula (8):Wherein slFor the standard variance of l item index in decision matrix Y,For l item index
Mathematic expectaion, ωlIt is l-th evaluation index weight coefficient in moment point t.
N number of weather region is pressed to the comprehensive evaluation index F in each region in moment point tAL,tPriority row from small to large
Sequence, and with comprehensive evaluation index FAL,tMinimum highest priority.
(q is between 1 to N to the weather region number q selecting in setting G (1≤G≤N) individual prediction scheme and each prediction scheme
Numerical value), and the numerical value of the q of each prediction scheme is different.For prediction scheme g, select moment point t priority higher
Q weather region, with the whole network system loading of selected q weather region difference prediction time point t, obtains prediction scheme g
Q different predicting the outcome, be expressed as:
Wherein,
It is the whole network system loading of the 1st moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of the 2nd moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of moment point t being predicted using q-th weather region of prediction scheme g,
AndLq,tFor q-th weather region moment point t predicted load, Cq,tFor q-th gas
As region is in the proportionality coefficient predicted value of moment point t.
The optimal synthesis model of moment point t is set up in q different the predicting the outcome to prediction scheme g, calculates selected
The optimal weights of each weather region, obtain moment point t the whole network system loading finally predict the outcome for
Wherein,It is the whole network system loading of being predicted by the weather region k of prediction scheme g optimum in moment point t
Weight,The whole network system loading of moment point t predicting out with k-th weather region for prediction scheme g;;
The moment point t optimal synthesis model of prediction scheme g is set up as follows:
Characterize the object function of moment point t the whole network system loading predicted value with formula (9), formula (10)-(11) are object function
Constraints:
In formula (9):Represent that the whole network system that jth sky moment point t of prediction scheme g is predicted by weather region k is born
Lotus predicted value, Lall,t,jRepresent the whole network system loading actual value of jth sky moment point t.
For the solution of formula (9)-(11), define the whole network system loading predicted value that each weather region predicts first
Virtual prognostication residual error vkjt, virtual prognostication residual sum of squares (RSS)And certain two weather region predicts the virtual of the whole network system loading
The covariance predicting the outcomeAs follows:
Then formula (9)-(11) object function in the t period is converted into following matrix form:
Wherein:
Formula (12) is the canonical form of quadratic programming problem, directly calculates the selected q meteorologic district of prediction scheme g
Domain is in the optimal weights of moment point tAfterwards, further according toWeighting obtains prediction side
The whole network system loading predicted value of case g moment point t.
For the whole network system loading that each weather region predicted load predicts, present in different moment point
Go out different prediction effects, therefore treat " each moment point of day to be predicted " with a certain discrimination, set up collective model respectively so that each
The whole network system loading that weather region predicted load predicts is all different in the weight in each moment, each comfortable different to embody
The prediction effect of moment point.For day to be predicted whole day T moment point, set up optimal synthesis model respectively, obtain prediction scheme g
Whole day load prediction sequence
Similarly, G whole day load prediction sequence, moment point t of each prediction scheme will be had for G prediction scheme
Optimal synthesis model is set up as follows:
Characterize the object function of moment point t the whole network system loading predicted value with formula (13), formula (14)-(15) are object function
Constraints:
In formula (13):Represent the whole network system loading predicted value that jth sky moment point t is predicted by prediction scheme g,
Lall,t,jRepresent the whole network system loading actual value of jth sky moment point t.
The optimal weights in moment point t for the G weather region are calculated by formula (13), formula (14) and formula (15)Afterwards, then
According toWeighting obtains the final predicted value of the whole network system loading of day to be predicted moment point t.
For the solution of formula (13)-(15), the solution with formula (9)-(11) is identical, same definition the whole network system loading
The virtual prognostication residual error of predicted value, virtual prognostication residual sum of squares (RSS), and prediction the whole network system loading of certain two prediction scheme
The covariance of virtual prognostication result, sets up such as the quadratic programming of formula (12), directly calculates prediction scheme g in moment point t
Optimal weightsAfterwards, further according toThe whole network system loading weighting moment point t obtaining day to be predicted is finally pre-
Measured value.For day to be predicted whole day T moment point, set up the optimal synthesis model of each prediction scheme respectively, obtain prediction side
Case g whole day load prediction sequence (L1,L2,…,LT), with described whole day load prediction sequence (L1,L2,…,LT) it is the whole network load
Predict the outcome.
Below with specific example, go on to say implementation process of the present invention:
In the methods of the invention, the meteorology letter to each administrative region using the synchronous back substitution technology for eliminating based on probability metrics
Breath is polymerized, and is divided into the weather region that N number of meteorological condition differs greatly.G prediction scheme of setting and each Forecasting Methodology
Selection weather region number q.Further according to the comprehensive evaluation index of each weather region, select and evaluate q optimum meteorologic district
Domain.Taking predict that certain saves the whole network load as a example, this province includes 16 districts and cities and (uses districts and cities 1, districts and cities 2 ... respectively, districts and cities 16 represent, ground
City's numbering criterion is:So that from north orientation south, geographical position from east to west is numbered to each districts and cities), according to meteorological condition phase
Like degree determination methods, 16 districts and cities are divided into the weather region that 6 meteorological conditions differ greatly, select 6 prediction scheme, often
The weather region number of individual prediction scheme corresponding selection is respectively 1,2,3,4,5,6, is embodied as carrying out according to the following steps:
1st, digital independent:Selection on May 18th, 2016 is day to be predicted, obtains prediction 30 working day the whole networks a few days ago respectively
And in the actual load data of prefectures and cities, prefectures and cities' temperature record of first 30 days (highest temperature, the lowest temperature) and prefectures and cities
The daily load prediction value to be predicted of report.
2nd, class weather region divides:Using the meteorology letter to 16 districts and cities for the synchronous back substitution technology for eliminating based on probability metrics
Breath is polymerized, and forms the weather region that N=6 meteorological condition differs greatly.Through synchronous back substitution technology for eliminating, DA=[
Districts and cities 14 of 10 districts and cities of 8 districts and cities of 3 districts and cities of 2 districts and cities of city 11], the diagonal element of 16 × 16 matrix D B is:
(1,1) | (2,2) | (3,3) | (4,4) | (5,5) | (6,6) | (7,7) | (8,8) |
2 | 0 | 0 | 1 | 7 | 8 | 3 | 0 |
(9,9) | (10,10) | (11,11) | (12,12) | (13,13) | (14,14) | (15,15) | (16,16) |
8 | 0 | 0 | 11 | 10 | 0 | 13 | 14 |
Therefore 6 weather region of formation are as shown in the table:
3rd, the multiple criteria system solves:Three single evaluation indexs are respectively:Certain weather region load is in moment point t
Comprehensive stability degree evaluation index F1,t, certain weather region load is in prediction synthesis accuracy rate evaluation index F of moment point t2,t, certain gas
As region load proportion coefficient is in stability evaluation index F of moment point t3,t.
Following table be in implementation steps 26 weather region in three evaluation index results of t=1 moment point.
Region | F1,t | F2,t | F3,t |
Weather region 1 | 0.3431 | 0.3779 | 0.0381 |
Weather region 2 | 0.1949 | 0.1781 | 0.0349 |
Weather region 3 | 0.1998 | 0.0930 | 0.0239 |
Weather region 4 | 0.3367 | 0.2099 | 0.0294 |
Weather region 5 | 0.1605 | 0.1266 | 0.0328 |
Weather region 6 | 0.9666 | 0.4578 | 0.0555 |
4th, build comprehensive evaluation index FAL,t:(F as described in description of the invention, first to 6 weather region1,t,
F2,t,F3,t) dimensionless process, obtain decision matrix Y, then seek the variation weights omega=(ω of each single index1,ω2,
ω3), finally solve comprehensive evaluation index FAL,t=ω × Y:
By calculating, this 6 weather region are after the comprehensive evaluation index sort result of t=1 moment point:
5th, prediction scheme is set:According to 6 weather region, without loss of generality, 6 prediction scheme, each prediction side are set
Case corresponding weather region number is respectively 1,2,3,4,5,6.
6th, choose weather region and predict the whole network load respectively:Here taking prediction scheme 3 as a example, its weather region number q=3.
For comprehensive evaluation index FAL,t, its numerical value is less, illustrates that the load of this weather region is more stable, is more conducive to using this meteorology
The predicted load in region is predicting the whole network load.Here according to FAL,tRanking results, corresponding respectively is weather region 3, gas
As region 5, weather region 2.ByThe 3 the whole network load prediction results obtaining in t=1 moment point are as follows:
7th, the optimal synthesis model of single moment point:The whole network predicted load of prediction t=1 moment point.To formula (9)-(11)
Solved, solving the optimal weights obtaining is:
The whole network predicted load of the therefore t=1 moment point of prediction scheme 3
And be 15069.9MW in the actual load of May t=1 moment point on the 18th, the precision of prediction of the t=1 moment point of prediction scheme 3 reaches
To 98.83%.
8th, the load prediction of whole day multiple spot.For t in one day day of prediction from 2 to 96 other load predictions put, repeat
Step 1 is to step 6 it is possible to obtain the load prediction value sequence of whole day.
9th, the load prediction of the whole day T moment point of each prediction scheme:Similar to step 6, step 7, step 8, each
Prediction scheme all carries out load prediction to 96 moment of one day day to be predicted, obtains the whole day predicted load of each prediction scheme
Sequence.
10th, the optimal synthesis model of each prediction scheme:Each prediction scheme is set up respectively for each moment point
Excellent collective model, solves to the planning problem of formula (13)-(15), obtains the optimum in t=1 moment point for each prediction scheme
Weight is:
Final the whole network predicted load of t=1 moment point therefore is:
Fig. 3 is that this saves 6 prediction scheme of on May 18th, 2016 and predicts the 6 the whole network load prediction curves obtaining respectively, figure
4 is load prediction curve and the realized load curve contrast of the final whole day through each prediction scheme optimal synthesis model solution
Figure.By calculating, the day accuracy rate of the inventive method is 98.87%.
General principle, principal character and the advantages of the present invention of the present invention have been shown and described above.The technology of the industry
, it should be appreciated that the present invention is not restricted to the described embodiments, the simply explanation described in above-described embodiment and specification is originally for personnel
Invention principle, without departing from the spirit and scope of the present invention the present invention also have various changes and modifications, these change
Change and improvement both falls within scope of the claimed invention.Claimed scope by appending claims and its
Equivalent defines.
Claims (10)
1. the bulk power grid load forecasting method based on class meteorology subregion overall merit is it is characterised in that comprise the steps:
(1) obtain the predicted load of the M administrative region having reported in daily load prediction value to be predicted;Obtain a recent sample
Historical data in this period, the load data in described historical data is the actual load and in advance of the whole network and M administrative region
Survey load;
(2) at least 2 history meteorological condition data in M administrative region in the sample period are obtained, by history meteorological condition
Data, is divided into the weather region that N number of meteorological condition differs greatly;
(3) calculate the average proportions coefficient in moment point t for the described N number of weather regionTreated pre- with exponential smoothing dynamic prediction
Survey day N number of weather region in the proportionality coefficient of identical moment point t, obtain the proportionality coefficient matrix in moment point t for N number of weather region
Ct;
(4) by the weight of multiple single evaluation indexs and each single evaluation index, build moment point t comprehensive evaluation index
FAL,t;
(5) N number of weather region is pressed to the comprehensive evaluation index F in each region in moment point tAL,tPrioritization from small to large,
And with comprehensive evaluation index FAL,tMinimum highest priority;
(6) the weather region number q selecting in G prediction scheme and each prediction scheme is set, and the number of the q of each prediction scheme
Value is different, and described G meets 1≤G≤N, and described q meets 1≤q≤N;
(7) for certain prediction scheme g, select q higher weather region of moment point t priority, meteorological with selected q
The whole network system loading of region difference prediction time point t, obtains q different predicting the outcome;
(8) q different predicting the outcome is set up with the optimal synthesis model of moment point t, calculates selected each weather region
Optimal weights, obtain moment point t the whole network system loading finally predict the outcome for
Wherein,It is the whole network system loading of being predicted by the weather region k of prediction scheme g optimal weights in moment point t,The whole network system loading of moment point t predicting out with k-th weather region for prediction scheme g;
(9) for day to be predicted whole day T moment point, set up optimal synthesis model respectively, obtain the whole day load of prediction scheme g
Forecasting sequence
(10) for different the predicting the outcome of G prediction scheme G, set up the optimal synthesis model of whole day T moment point respectively,
Obtain the final load prediction results (L of day to be predicted1,L2,…,LT):
Wherein,The whole network system loading predicting for prediction scheme g is in the optimal weights of moment point t.
2. the bulk power grid load forecasting method based on class meteorology subregion overall merit according to claim 1, its feature exists
In:Described at least 2 history meteorological condition data are in the sample period, the maximum temperature of administrative region, minimum temperature, flat
Any group of in equal temperature, human comfort, weather pattern, wind speed, comfort index, coldness index 2 or more than 2
Close.
3. the bulk power grid load forecasting method based on class meteorology subregion overall merit according to claim 1, its feature exists
In:Described at least 2 history meteorological condition data be a sample period in, the maximum temperature of administrative region, minimum temperature and
Mean temperature, the history meteorological condition data according to M administrative region forms meteorological data matrix X:
X=(X1,X2,…Xi,…,XM)
Wherein,
XiFor i-th administrative region within a sample period meteorological condition data acquisition system of n days,
For i-th administrative region within a sample period maximum temperature of n days;
For i-th administrative region within a sample period minimum temperature of n days;
For i-th administrative region within a sample period mean temperature of n days.
4. the bulk power grid load forecasting method based on class meteorology subregion overall merit according to claim 3, its feature exists
In:Using the synchronous back substitution technology for eliminating based on probability metrics, matrix X is polymerized, is divided into N number of meteorological condition difference relatively
Big weather region:
First, a Probability p occurring is specified to M column vector in matrix Xs=1/M (s=1,2 ..., M).Set S represents
First administrative region, makes ξsS-th administrative region in representing matrix X, DTs,s'Represent administrative region s and administrative region s' away from
From its value is vector 2 norm between administrative region s and administrative region s', the vectorial DA=[1,2 ..., M] of settingT, DB is institute
There is the square formation that element is all 0 M × M.
The basic step that synchronous back substitution eliminates is as follows:
1) calculate the distance between each pair administrative region DTs,s', DTs,s'=| | ξs,ξs'||.
2) for each administrative region k, find out the administrative region r, i.e. DT the shortest with administrative region k distancek(r)=min
DTk,s'.
3) calculate PDk(r)=pk*DTk(r), k ∈ S, finds out administrative region index h so that PD in kh=min PDk,k∈S.
4) make S=S- ξDA(h), and pr=pr+ph.DB (h, h)=DA (r), DA (h) are set to sky.
5) 2 are repeated) -4), till remaining administrative region number is N.
5. according to the arbitrary described bulk power grid load forecasting method based on class meteorology subregion overall merit of Claims 1-4, its
It is characterised by:In described step (1), described historical data is pre-processed as follows:
Order:L (d, t) is the load value of the d days t, L (d, t1) and L (d, t2) it is the d days two adjacent with t moment
t1、t2Load value, L (d1, t) with L (d2, t) it is the load value in moment point t for two adjacent with d day;
A) for the process of missing data
If load value L (d, the t) disappearance of the d days t, then formula (1) is utilized to obtain L (d, t):
L (d, t)=α L (d, t1)+αL(d,t2)+βL(d1,t)+βL(d2,t) (1)
In formula (1), α and β is coefficient, α>β, alpha+beta=1;
B) for the process of bad point data
Define the bias ratio that allows for load of ε, ρ (d, t) is the actual bias ratio of the d days moment point t, as ρ (d, t) >=ε, judgement
L (d, t) is bad data, and bad data is usedReplaced:
6. according to the arbitrary described bulk power grid load forecasting method based on class meteorology subregion overall merit of Claims 1-4, its
It is characterised by:Described step (3) is expressed as:
Ct=(C1,t,C2,t,…,CN,t)
Wherein,
For the 1st weather region historical data moment point t average proportions coefficient;
For the 2nd weather region historical data moment point t average proportions coefficient;
For n-th weather region historical data moment point t average proportions coefficient;
C1,tFor the 1st weather region day to be predicted moment point t proportionality coefficient;
C2,tFor the 2nd weather region day to be predicted moment point t proportionality coefficient;
CN,tFor n-th weather region day to be predicted moment point t proportionality coefficient.
7. the bulk power grid load forecasting method based on class meteorology subregion overall merit according to claim 6, its feature exists
In:Exponential smoothing in described step (3) is that onset index smoothing model is:
In formula (3):Represent that weather region i accounts for the predicted value of the whole network system loading ratio in moment t,J days before expression
Weather region i accounts for the actual value of the whole network system loading ratio in moment t;N is the number of days within a sample period;λjRepresent power
Weight coefficient, λj=λ (1- λ)j-1, λ is constant, and 0<λ<1.
8. according to the arbitrary described bulk power grid load forecasting method based on class meteorology subregion overall merit of Claims 1-4, its
It is characterised by:In described step (4), build multiple single evaluation index systems F of moment point tt, referred to using three single evaluation
Mark, then the multiple criteria system F of each weather regiontFor:
Ft=(F1,t,F2,t,F3,t)T
Described three single evaluation indexs are respectively:Certain weather region load is in the comprehensive stability degree evaluation index of moment point t
F1,t, certain weather region load is in prediction synthesis accuracy rate evaluation index F of moment point t2,t, certain weather region load proportion coefficient
Stability evaluation index F in moment point t3,t;
Build moment point t comprehensive evaluation index FAL,tFor;
FAL,t=ω × Y
Wherein:ω=(ω1,ω2,ω3), ω is the weight matrix in moment point t for three single evaluation indexs;ω1,ω2,ω3
For multiple single evaluation index systems FtIn each single evaluation index moment point t weight coefficient;Y is according to multiple single
Assessment indicator system FtThe decision matrix in moment point t being formed.
9. the bulk power grid load forecasting method based on class meteorology subregion overall merit according to claim 8, its feature exists
In:Certain weather region load described is in comprehensive stability degree evaluation index F of moment point t1,tObtained by formula (4):
In formula (4):RSDtFor certain weather region moment t weather region load relative standard deviation,Meteorological for certain
Region accounts for the average proportions coefficient of system loading in moment t time domain load;
Certain weather region load described is in prediction synthesis accuracy rate evaluation index F of moment point t2,tObtain by formula (5):
In formula (5):QtConsensus forecast accuracy rate for moment t in selected sample space for certain weather region;
Certain weather region load proportion coefficient described is in stability evaluation index F of moment point t3,tObtain by formula (6):
In formula (6):For the sample canonical variance of the proportionality coefficient in moment t for certain weather region,For certain meteorologic district
The sample mathematic expectaion of the proportionality coefficient in moment t for the domain;
Or, each single evaluation index described is in the weight coefficient ω of moment point t1,ω2,ω3And decision matrix Y is by as follows
Method obtains:
1) the decision matrix Y is made to be:Y=(yil)3×N, wherein:
In formula (7):Fi,lL-th evaluation index for weather region i, mini{Fi,lIt is evaluating for l-th in N number of weather region
The minimum of a value of index, maxi{Fi,lBe l-th evaluation index in N number of weather region maximum;
2) then have:
In formula (8):Wherein slFor the standard variance of l item index in decision matrix Y,Number for l item index
Term hopes, ωlIt is l-th evaluation index weight coefficient in moment point t.
10. according to the arbitrary described bulk power grid load forecasting method based on class meteorology subregion overall merit of Claims 1-4,
It is characterized in that:In described step (7), for certain prediction scheme g, obtain q different predicting the outcome, be expressed as:
Wherein,
It is the whole network system loading of the 1st moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of the 2nd moment point t that weather region predicts using prediction scheme g;
It is the whole network system loading of moment point t being predicted using q-th weather region of prediction scheme g, andLq,tFor q-th weather region moment point t predicted load, Cq,tFor q-th weather region when
The proportionality coefficient predicted value of punctum t;
In described step (8), optimal synthesis model is set up as follows:
Characterize the object function of moment point t the whole network system loading predicted value with formula (9), formula (10)-(11) are the pact of object function
Bundle condition:
In formula (9):Represent that the whole network system loading that jth sky moment point t of prediction scheme g is predicted by weather region k is pre-
Measured value, Lall,t,jRepresent the whole network system loading actual value of jth sky moment point t;
Q weather region selected in prediction scheme g is calculated in moment point t by formula (9), formula (10) and formula (11)
Excellent weightAfterwards, further according toThe whole network system loading weighting moment point t obtaining prediction scheme g is pre-
Measured value;
In described step (10), G prediction scheme will obtain G different predicting the outcome, and optimal synthesis model is as follows
Set up:
Characterize the object function of moment point t the whole network system loading predicted value with formula (12), formula (13)-(14) are the pact of object function
Bundle condition:
In formula (12):Represent the whole network system loading predicted value that jth sky moment point t is predicted, L by prediction scheme gall,t,j
Represent the whole network system loading actual value of jth sky moment point t.
The optimal weights in moment point t for the G weather region are calculated by formula (12), formula (13) and formula (14)Afterwards, further according toWeighting obtains the final predicted value of the whole network system loading of day to be predicted moment point t.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610815916.1A CN106408119A (en) | 2016-09-09 | 2016-09-09 | Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610815916.1A CN106408119A (en) | 2016-09-09 | 2016-09-09 | Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106408119A true CN106408119A (en) | 2017-02-15 |
Family
ID=57999080
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610815916.1A Pending CN106408119A (en) | 2016-09-09 | 2016-09-09 | Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106408119A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107506843A (en) * | 2017-07-03 | 2017-12-22 | 国网上海市电力公司 | A kind of short-term load forecasting method and device |
CN108229754A (en) * | 2018-01-31 | 2018-06-29 | 杭州电子科技大学 | Short-term load forecasting method based on similar day segmentation and LM-BP networks |
CN109861222A (en) * | 2019-03-29 | 2019-06-07 | 国网湖南省电力有限公司 | A kind of provincial power network overloaded partition prediction technique and system |
CN112036004A (en) * | 2020-07-14 | 2020-12-04 | 国网江苏省电力有限公司检修分公司 | Method and system for predicting oil temperature of phase modifier oil system based on similar time dynamic selection |
CN113222230A (en) * | 2021-04-29 | 2021-08-06 | 中国石油大学(北京) | Flow distribution method and device of natural gas pipe network under accident condition |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103093288A (en) * | 2013-02-21 | 2013-05-08 | 江苏省电力公司电力科学研究院 | Partition power grid bus load prediction system based on weather information |
CN103617564A (en) * | 2013-12-04 | 2014-03-05 | 国家电网公司 | Whole-network load prediction method based on local load predicted value comprehensive evaluation |
-
2016
- 2016-09-09 CN CN201610815916.1A patent/CN106408119A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103093288A (en) * | 2013-02-21 | 2013-05-08 | 江苏省电力公司电力科学研究院 | Partition power grid bus load prediction system based on weather information |
CN103617564A (en) * | 2013-12-04 | 2014-03-05 | 国家电网公司 | Whole-network load prediction method based on local load predicted value comprehensive evaluation |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107506843A (en) * | 2017-07-03 | 2017-12-22 | 国网上海市电力公司 | A kind of short-term load forecasting method and device |
CN108229754A (en) * | 2018-01-31 | 2018-06-29 | 杭州电子科技大学 | Short-term load forecasting method based on similar day segmentation and LM-BP networks |
CN108229754B (en) * | 2018-01-31 | 2021-12-10 | 杭州电子科技大学 | Short-term load prediction method based on similar day segmentation and LM-BP network |
CN109861222A (en) * | 2019-03-29 | 2019-06-07 | 国网湖南省电力有限公司 | A kind of provincial power network overloaded partition prediction technique and system |
CN109861222B (en) * | 2019-03-29 | 2022-12-06 | 国网湖南省电力有限公司 | Provincial power grid load partition prediction method and system |
CN112036004A (en) * | 2020-07-14 | 2020-12-04 | 国网江苏省电力有限公司检修分公司 | Method and system for predicting oil temperature of phase modifier oil system based on similar time dynamic selection |
CN112036004B (en) * | 2020-07-14 | 2024-04-19 | 国网江苏省电力有限公司检修分公司 | Method and system for predicting oil temperature of phase modulation engine oil system based on dynamic selection of similar moments |
CN113222230A (en) * | 2021-04-29 | 2021-08-06 | 中国石油大学(北京) | Flow distribution method and device of natural gas pipe network under accident condition |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106408119A (en) | Large power grid load prediction method based on comprehensive evaluation of analog-meteorological subareas | |
CN103617564B (en) | The whole network load forecasting method based on region predicted load overall merit | |
CN108230049A (en) | The Forecasting Methodology and system of order | |
CN109558975B (en) | Integration method for multiple prediction results of power load probability density | |
CN102426674B (en) | Power system load prediction method based on Markov chain | |
CN109214581A (en) | A kind of Along Railway wind speed forecasting method considering wind direction and confidence interval | |
CN103218675A (en) | Short-term load prediction method based on clustering and sliding window | |
CN102509173B (en) | A kind of based on markovian power system load Accurate Prediction method | |
CN107437135B (en) | Novel energy storage type selection method | |
CN108664682A (en) | A kind of prediction technique and its system of transformer top-oil temperature | |
CN104933627A (en) | Energy efficiency combination evaluation method of machine tool product manufacture system | |
CN108695902A (en) | A kind of Cascade Reservoirs ecology-mutual feedback regulation and control method of power generation dynamic | |
CN102509240A (en) | Grid investment benefit evaluation method based on multiple indexes and multiple levels | |
CN106803129A (en) | A kind of wind power ensemble prediction method based on multi-source numerical weather forecast | |
CN103942422B (en) | Granular-computation-based long-term prediction method for converter gas holder positions in metallurgy industry | |
CN106447091A (en) | Regional meteorological condition similarity-based large power network load prediction method | |
CN114036452B (en) | Yield evaluation method applied to discrete production line | |
CN107358332A (en) | A kind of dispatching of power netwoks runs lean evaluation method | |
CN109523077B (en) | Wind power prediction method | |
CN105846425A (en) | Economic dispatching method based on general wind power forecasting error model | |
CN107220735A (en) | A kind of multivariable rural power grids power predicating method of power industry classification | |
CN107358363A (en) | Coal work incidence of disease Forecasting Methodology based on radial basis function neural network built-up pattern | |
CN105701562A (en) | Training method, suitable method of predicating generated power and respective systems | |
CN105303268A (en) | Wind power generation output power prediction method based on similarity theory | |
CN104504280A (en) | Planning-demand-considered comprehensive evaluation method for communication mode of cluster management system of charging piles of electric automobile |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170215 |