CN106600449B - Automatic power trend identification method - Google Patents
Automatic power trend identification method Download PDFInfo
- Publication number
- CN106600449B CN106600449B CN201510673427.2A CN201510673427A CN106600449B CN 106600449 B CN106600449 B CN 106600449B CN 201510673427 A CN201510673427 A CN 201510673427A CN 106600449 B CN106600449 B CN 106600449B
- Authority
- CN
- China
- Prior art keywords
- data
- active power
- trend
- value
- oscillation
- 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 39
- 230000010355 oscillation Effects 0.000 claims abstract description 63
- 238000007781 pre-processing Methods 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims abstract description 27
- 238000009499 grossing Methods 0.000 claims abstract description 25
- 238000004364 calculation method Methods 0.000 claims abstract description 11
- 238000013519 translation Methods 0.000 claims description 13
- AYFVYJQAPQTCCC-GBXIJSLDSA-N L-threonine Chemical compound C[C@@H](O)[C@H](N)C(O)=O AYFVYJQAPQTCCC-GBXIJSLDSA-N 0.000 claims description 6
- 238000003491 array Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000007774 longterm Effects 0.000 abstract description 3
- 230000008859 change Effects 0.000 description 8
- 230000002159 abnormal effect Effects 0.000 description 7
- 238000005259 measurement Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 3
- 230000001360 synchronised effect Effects 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Landscapes
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to an automatic power trend identification method, which comprises the following steps: the server receives active power original data in real time; preprocessing original data; active power oscillation state identification; and identifying active power trend. The technical scheme provided by the invention adopts a high-frequency data smoothing processing technology, and combines the experience of people in a long-term calculation process and a specific calculation method to realize the automatic identification of the power trend and the oscillation so as to achieve the purposes of assisting the signal trend and the oscillation identification of the power system and guiding the actual operation.
Description
Technical Field
The invention relates to an identification method of an electric power system, in particular to an automatic power trend identification method.
Background
In the operation process of the power system, the power system is inevitably influenced by hidden dangers of natural environment, human, communication or system internal elements. After the power system disturbance occurs, the system disturbance scene is timely and accurately acquired, and the disturbance type and position are accurately identified, so that reasonable and effective control measures can be implemented, and the safe and stable operation of a power grid is guaranteed. The WAMS is a new generation of whole-network monitoring system which is a basic element of a synchronous phasor measurement unit PMU, and can realize synchronous real-time acquisition and remote real-time transmission of whole-network data, so that the whole-network dynamic monitoring becomes possible. When disturbance occurs, PMUs distributed all over the network send acquired information to a scheduling center in real time, and operating personnel process and screen real-time synchronous data to obtain effective information representing disturbance characteristics and classify the disturbance to determine the next countermeasure. However, it is still difficult to automatically recognize the occurrence of a disturbance. In the existing technical scheme for identifying disturbance occurrence, a threshold value method is mainly adopted for judgment.
The calculation formula of power fluctuation in the existing method is as follows:
in the above formula, Δ P represents the degree of change of the power after disturbance relative to the normal power before disturbance, and Δ P is positive when the power is increased and negative when the power is decreased.
By setting different thresholds for the different measurement values, the system operation is considered to be disturbed when the observed quantity exceeds the threshold. Although the above technical solution has a fast calculation speed, it also has the following disadvantages:
(1) hard thresholding sometimes results in false positives due to noise interference of PMU measurement information, sometimes the measurement value exceeds the threshold value only because of noise interference.
(2) Reasonable threshold setting is very difficult. Since the threshold value should be adjusted accordingly as the grid operation is adjusted.
(3) Each measured value of PMU in the power grid is the expression of disturbance characteristics, and the same kind of electrical quantities measured by different PMU substations are influenced differently by disturbance due to different positions relative to the disturbance occurrence positions. The closer to the disturbing electrical distance is affected more and the farther away is affected less. And the disturbance occurrence position is uncertain, so the existing threshold disturbance judgment method is unreasonable.
Disclosure of Invention
The invention provides an automatic power trend identification method aiming at the problems of oscillation identification and trend judgment based on power grid data.
The purpose of the invention is realized by adopting the following technical scheme:
the invention provides 1 an automatic power trend identification method, which is improved in that the method comprises the following steps:
(1) the server receives active power original data in real time;
(2) preprocessing original data;
(3) active power oscillation state identification;
(4) and identifying active power trend.
Further, in the step (2), a moving average smoothing mode is adopted for preprocessing the original data to achieve data smoothing, and the active power change trend is extracted; the raw data preprocessing comprises the following steps:
1) extracting average values at equal intervals from the raw data, including: inputting original data Dt, extracting an average value of each T points of the original data Dt, recording the average value as 1 point, and storing the 1 point in an A variable; the A variables include A (1), A (2), A (3), A (4.); the A (1), A (2), A (3) and A (4) are Dt (1: T), Dt (T + 1: 2T), Dt (2T + 1: 3T) and Dt (3T + 1: 4T) respectively;
2) moving average processing is carried out on the extracted average value data so as to achieve the purpose of smoothing the data, and the method comprises the following steps: carrying out moving average processing on the variable A by adopting the following formula;
wherein: b is the processed data, i represents the data subscript parameter, j is the cycle parameter of the accumulated sum, D represents the length of the translation interval;
let the length of the data a be Len, and N be the result of Len dividing D, then the value range of the subscript parameter i is 1,2, …, N;
3) expanding the data after the moving smoothing, comprising: expanding each data point in the data B by 24 values forwards, wherein the expanded value is the value of the current point, and the expanded value is recorded as a variable C after the expansion; the C variables include C (1: T), C (T + 1: 2T), C (2T + 1: 3T), C (3T + 1: 4T).
Further, the step (3) comprises the steps of:
initializing the following parameters according to different unit characteristics, wherein the parameters comprise:
A. active power oscillation threshold percentage Per; the parameter is used for adjusting the sensitivity of oscillation feature identification, the larger the parameter value is, the smaller the oscillation identification area is, and otherwise, the larger the oscillation identification area is;
B. adding a footmark record amount U _ ind; the system comprises a data processing unit, a data processing unit and a data processing unit, wherein the data processing unit is used for recording the positions of all the subscripts of the maximum extreme points in original data;
C. the lower subscript record amount D _ ind; the system comprises a data processing unit, a data processing unit and a data processing unit, wherein the data processing unit is used for recording the positions of all minimum extreme points in original data;
smoothly preprocessing the original active power data by using the preprocessing steps 1) and 2) in the step 2, and returning the unexpanded preprocessed active power data;
searching an upper envelope Up and a lower envelope Down in the oscillation state;
fourthly, performing linear interpolation on the found upper envelope data Up and lower envelope data Down to obtain interpolated data U _ A and D _ A; the linear interpolation algorithm is as follows:
for the envelope data Up, first two data point coordinate information (x1, y1) and (x2, y2) are taken; wherein, the x coordinate represents the time, and the y coordinate represents the active power value of the corresponding time; if x2-x 1> 1, then linear interpolation is performed on the point x between x1 and x2 to obtain the corresponding y value, as follows:
y＝(y2-y1)/(x2-x1)*(x-x1)+y1
then, linear interpolation is carried out on (x2, y2) and (x3, y3), and the U _ A is obtained after the steps are sequentially carried out; in the same way, D _ A is obtained from the lower envelope data Down;
smoothing the interpolated upper envelope line data U _ A, interpolated lower envelope line data D _ A and original data by respectively reusing the algorithm in the preprocessing step, and marking the finally obtained data as Temp _ U, Temp _ D and M;
sixthly, detecting the oscillation state of the active power according to the oscillation judgment rule of the active power; the active power oscillation judgment rule is as follows: when min ((Temp _ U-M), (M-Temp _ D)) > M × Per, judging that the active power oscillates, and if not, ending;
and seventhly, expanding the judgment result to achieve the aim of matching with the original data quantity.
Further, the active power oscillation threshold percentage Per is between 0.005 and 0.01.
Further, finding the upper envelope when oscillating includes: (a) selecting continuous d data from the starting point of the active power data, searching the maximum data and the corresponding upper foot mark in the d data, selecting the data with the large upper foot mark if two or more maximum data exist, respectively storing the maximum data and the corresponding upper foot mark in arrays Up and U _ ind, and marking the foot mark as a variable index; (b) repeating the steps (a) and (b) by taking index +1 as a starting point until all data are inquired and the searching of the upper envelope line in the oscillation state is finished;
finding the lower envelope when oscillating includes: (c) selecting continuous D data from the starting point of the data, finding out the minimum data and the corresponding subscript mark in the D data, if two or more than two minimum data exist, selecting the data with the small subscript mark, respectively storing the minimum data and the corresponding subscript mark in arrays Down and D _ ind, and marking the subscript mark as index' (); (d) and (d) repeating the steps (c) and (d) by taking index' +1 as a starting point until all data are inquired and the searching of the lower envelope line in the oscillation state is finished.
Further, the step (4) comprises the steps of:
initializing parameters including a historical data segment k, a smooth data interval parameter D, an active power trend threshold percentage Per and a translation amount py according to different unit characteristics;
receiving 25 × D unprocessed active power signal sampling values, performing downward translation and smoothing on active power data by using steps 1) and 2) of a preprocessing method, recording as B, and returning unexpanded data;
<3>according to the active power preprocessing data, carrying out weighted average on the first k data of the current data B (j) to obtain a comprehensive index
Wherein the weight satisfies w (j-k) is less than or equal to w (j- (k-1)). less than or equal to w (j-1), and the threshold value thre ═ C × Per; j is the subscript of the current processing data segment, m is a cycle parameter, w (j-m) is a corresponding weight parameter, and B (k-m +1) represents the value of an array B corresponding to the subscript;
combining the calculation result of the trend judgment comprehensive index, and detecting active power trend judgment by utilizing a trend judgment rule;
<5> executing a trend extension strategy if the active power trend is an ascending trend or a descending trend according to the preliminary judgment result;
and 6, expanding the judgment result, namely expanding each data point in the B data forward by 24, wherein the expanded value is the numerical value of the current point, and the expanded state is the state of the current point.
Further, in the step <4>, the detecting of the active power trend judgment includes:
judging a trend judgment rule abs (B (j) -C) > thre, if B (j) < C Down is satisfied, judging that the active power trend is an ascending trend, otherwise, judging that the active power trend is a descending trend; if abs (B (j) -C) < thre, judging that the trend of the active power is stable, and continuing to the next step, wherein Down represents lower envelope data; thre represents a threshold; c is a variable after the data after the movement smoothing is expanded; b (j) represents the current data.
Further, the value range of the historical data segment k is 8-12, and the value range of the smooth data volume D is 35-50; the active power trend threshold percentage Per is between 1 and 5; the value of the translation amount py is selected as the minimum value reached when the corresponding unit normally works.
The technical scheme provided by the invention has the following excellent effects:
1. when the system is influenced by disturbance, the occurrence of the disturbance can be found as soon as possible, which has important significance for subsequent system operation control and post-accident analysis. Under the normal operation condition, the method can automatically judge the starting time of the disturbance and can identify the power fluctuation trend and the oscillation of the system, which has important significance for ensuring the safety and the reliability of the operation of the power grid.
2. The method provided by the invention is based on real-time data of an actual power grid, researches and attempts for automatically identifying power oscillation and fluctuation trend identification methods are carried out, and the automatic trend identification method based on the PMU is provided by combining the accumulated trend observation experience of manual long-term research, a necessary calculation method and the characteristics of an actual power grid system. The method can automatically identify the oscillation and fluctuation characteristics of the observed quantity by utilizing a trend analysis technology, and has higher calculation speed. The trend identification method can be applied to an actual power grid and meets the requirement of rapidity of the actual power grid.
3. The method provided by the invention changes the traditional mode of starting disturbance judgment by means of a threshold value method, realizes online real-time calculation and intelligent trend identification, is a further improvement on the traditional disturbance starting judgment mode, provides an important guidance function for the actual power grid operation, and is beneficial to improving the safety of the actual power grid operation.
Drawings
FIG. 1 is a flow chart of a PMU measurement trend identification model provided by the present invention;
FIG. 2 is a flow chart of data preprocessing provided by the present invention;
FIG. 3 is a flow chart of active power oscillation identification provided by the present invention;
FIG. 4 is a flow chart of active power trend identification provided by the present invention;
FIG. 5 is a graph of active power oscillation identification provided by the present invention;
fig. 6 is an active power trend identification chart provided by the present invention.
Detailed Description
The following describes embodiments of the present invention in further detail with reference to the accompanying drawings.
The following description and the drawings sufficiently illustrate specific embodiments of the invention to enable those skilled in the art to practice them. Other embodiments may incorporate structural, logical, electrical, process, and other changes. The examples merely typify possible variations. Individual components and functions are optional unless explicitly required, and the sequence of operations may vary. Portions and features of some embodiments may be included in or substituted for those of others. The scope of embodiments of the invention encompasses the full ambit of the claims, as well as all available equivalents of the claims. Embodiments of the invention may be referred to herein, individually or collectively, by the term "invention" merely for convenience and without intending to voluntarily limit the scope of this application to any single invention or inventive concept if more than one is in fact disclosed.
The method is based on the power grid data, aims at the problems of oscillation identification and trend judgment, adopts a high-frequency data smoothing processing technology, combines the experience of people in the long-term calculation process and a specific calculation method, and realizes the automatic identification of the power trend and the oscillation so as to achieve the purposes of assisting the signal trend and the oscillation identification of the power system and guiding the actual operation.
Based on the characteristics of large data quantity and high frequency of PMU measurement values, in order to reduce the great trouble brought to a trend judgment program by the fluctuation of data, the original data is firstly smoothed to obtain preprocessed data. And then a corresponding trend and oscillation identification algorithm is provided for the power.
The preprocessing and power trend and oscillation identification are performed in 3 parts, and detailed description is divided into blocks.
(1) The server receives active power original data in real time;
(2) data preprocessing:
the data preprocessing process mainly carries out corresponding data processing before identifying the trend of the original signal. The main idea is as follows: because 50 data with constantly changing values are generated within 1s, the data fluctuation phenomenon is caused, and the data fluctuation amplitude is influenced by the data type and the data generation environment, a data preprocessing means is adopted. And in the preprocessing stage, a moving average smoothing method is adopted, so that the effects of smoothing data and extracting the overall trend are achieved. The algorithm comprises the following specific steps:
1) inputting data Dt, taking an average value of every T interval points of the data as 1 point, and storing the average value in an A variable;
2) carrying out moving average processing on the A by adopting the following formula, and setting the translation amount as D;
wherein: i is a subscript of the processed data B, j is a cycle parameter during accumulation and T represents an interval length;
3) and (3) expanding each data point in the B data by 24 forwards, wherein the expanded value is the numerical value of the current point. And after the expansion is finished, marking as a C variable, and returning to the C.
The specific execution flow of data preprocessing is shown in fig. 2.
(3) Active power oscillation identification: the method mainly comprises the following steps of carrying out intelligent analysis and automatic characteristic identification on an active power curve, wherein the method comprises the following steps: the method comprises the steps of detecting an oscillation starting point, detecting an oscillation ending point, detecting the change of an oscillation trend and the like, and providing the identification characteristics of active power through a self-adaptive algorithm.
General active power monitoring signals have fluctuation phenomena, and the fluctuation amplitude is closely related to a power grid system. Oscillation refers to the phenomenon that the power signal fluctuates to a larger extent in a sampling period. Typically the oscillation amplitude will be much larger than the normal fluctuation amplitude. The correct detection of the onset of the oscillation signal characteristic is the key to determining the onset of oscillation and is also the basis for determining the oscillation trend.
The oscillation identification of the active power of the power grid needs to be combined with the specific characteristics of the current power grid, and the normal fluctuation threshold value of the active power in the power grid is determined. And comparing the change value of the active power signal in unit time with a normal fluctuation threshold value to judge the generation of the oscillation signal. Thereby identifying the oscillation start point and the oscillation end point. When oscillation occurs, an envelope detection algorithm is used for giving the dynamic change trend of the oscillation range.
And utilizing the active power signal to automatically identify the oscillation characteristics. The specific algorithm steps are as follows:
1) initializing parameter threshold value percentage Per, upper foot mark recording amount U _ ind and lower foot mark recording amount D _ ind;
2) inputting data and preprocessing;
3) finding the upper envelope: d consecutive data are selected from the beginning of the data. Finding out the maximum data and the corresponding foot mark of the section, respectively storing the maximum data and the corresponding foot mark in the arrays Up and U _ ind, and marking the foot mark as index. (if there are two or more maximum data, then select the data with larger subscript)
4) Repeating the steps 3) and 4) by taking index +1 as a starting point until all data are inquired, and finishing searching the upper envelope line;
5) the lower envelope is sought: d consecutive data are selected from the beginning of the data. Finding out the minimum data and corresponding foot mark, respectively storing in arrays Down and D _ ind, and marking the foot mark as index (if there are two or more minimum data, selecting data with smaller foot mark);
6) repeating steps 5) and 6) starting from index + 1. Until all data are inquired, the searching of the lower envelope line is finished;
7) performing linear interpolation on Up and Down, and recording the interpolation as U _ A, D _ A; the linear interpolation algorithm is as follows:
for the envelope data Up, first two data point coordinate information (x1, y1) and (x2, y2) are taken; wherein, the x coordinate represents the time, and the y coordinate represents the active power value of the corresponding time; if x2-x 1> 1, then linear interpolation is performed on the point x between x1 and x2 to obtain the corresponding y value, as follows:
y＝(y2-y1)/(x2-x1)*(x-x1)+y1
then, linear interpolation is carried out on (x2, y2) and (x3, y3), and the U _ A is obtained after the steps are sequentially carried out; in the same way, D _ A is obtained from the lower envelope data Down;
8) respectively carrying out smoothing treatment on the interpolated upper envelope line data U _ A, interpolated lower envelope line data D _ A and original data by using the algorithm in the preprocessing step again, and finally recording the obtained data as Temp _ U, Temp _ D and M;
9) detecting the oscillation state of the active power according to the oscillation judgment rule of the active power; the active power oscillation judgment rule is as follows: when min ((Temp _ U-M), (M-Temp _ D)) > M × Per, judging that the active power oscillates, and if not, ending;
and expanding the judgment result to achieve the purpose of matching with the original data quantity.
The automatic identification result of the oscillation characteristic of the active power signal is shown in fig. 5, wherein the portion of the dashed line is an envelope and a smooth intermediate value, and the start point and the end point of the dashed line are the start point and the end point of the oscillation. If the threshold value is increased, the oscillation area is judged to be narrowed, namely, the broken line envelope part in the figure 5 is reduced; the threshold value is reduced, the starting point of oscillation can be accurately judged, and meanwhile, the state misjudgment can be caused. Therefore, the parameter values are set so that the result is shown in fig. 5, in consideration of the combination.
In the implementation, there are several parameters that affect the result of the algorithm.
1) A power normal fluctuation range threshold.
In actual judgment, the threshold is measured as a percentage of the smoothed average. The size of the percentage value is therefore dependent on the size of the indicator itself. Generally, the smaller the Per value, the more rising and falling portions are monitored, and the more smaller fluctuation anomaly intervals are detected. When the Per value is larger, the abnormal state is less detected, and the detection omission phenomenon may occur. Per values are generally recommended to be between 0.005 and 0.01.
The specific execution flow of the active power oscillation determination is shown in fig. 3.
(4) Active power trend identification:
besides the oscillation phenomenon, the active power signal is in a normal fluctuation range for more time. In the normal fluctuation range, three variation trends exist in the active power signal: normal, up and down. Since the trend detection is an analysis of the overall trend change of the active power signal, the signal needs to be smoothed to a larger scale before the trend analysis, so as to eliminate the influence of glitch signal noise and the like on the overall trend of the power signal. Because the power signals are not of uniform order, first, the original data needs to be shifted. Meanwhile, because the sampling frequency of the power signal is high (50Hz), 1 equivalent point is selected from every 25 points of the original data before the whole trend change is judged, and then the equivalent points are subjected to smoothing treatment. And when the trend changes, extracting indexes of the current data and the previous k continuous data for automatic trend identification.
And utilizing the active power signal to automatically identify the variation trend. The specific algorithm steps are as follows:
1) initializing parameters: relating to a historical data segment k, and smoothing an interval point D; a threshold percentage Per; translation amount py;
2) the input raw data is translated downwards and smoothed, and is marked as B. The smoothing method is shown in the data preprocessing steps (2) and (3); (Note that the power pre-processing stage does not perform step (4), but rather returns the smoothed, unamplified data directly).
3) Carrying out weighted average on the first k data of the current data B (j) to obtain an index
Wherein the weight satisfies w (j-k) is less than or equal to w (j- (k-1)). less than or equal to w (j-1); the threshold value thre ═ C × Per;
4) carrying out automatic trend recognition:
5) trend extension strategy: and 4, reasonably extending the abnormal state under the condition that the power fluctuation is judged to be abnormal in the step 4. Taking the descent as an example, by adjusting the value of the parameter (hupu _ down), the trend extension can be performed on the portion which is not accurately identified as the descent in step 4, but which is on the premise that the overall trend is the descent.
6) And (5) data expansion. And expanding each data point in the B data by 24 forwards, wherein the expanded value is the numerical value of the current point, and the expanded state is the state of the current point.
The specific execution flow of the active power trend determination is shown in fig. 4.
The automatic active power signal trend identification result is shown in fig. 6, and it is easy to see that the invention makes reasonable judgment on the abnormal states of the decline and the rise. The previous section of data judged to be in the descending trend is in a normal state, and whether the data is in the descending trend is hard to judge manually in real time judgment without combining the data at the future moment. The conventional methods described in the background art cannot necessarily achieve the effects of the present invention.
In the implementation, there are several parameters that affect the result of the algorithm.
1) And (4) a threshold value of the normal fluctuation range of the active power.
In actual judgment, the threshold is measured as a percentage of the smoothed average. The size of the percentage value is therefore dependent on the size of the data itself. Generally, the smaller the Per value, the more rising and falling portions are monitored, and the more smaller fluctuation anomaly intervals are detected. When the Per value is larger, the abnormal state is less detected, and the detection omission phenomenon may occur. Per values of 1 to 5 are generally suggested.
2) Smoothing data volume D
That is, D pieces of original data are extracted each time as data in one interval, and moving average processing is performed on the D pieces of data to achieve the purpose of smoothing the data. The value of D affects the accuracy of the determination result and the time delay. Generally, the larger the value of D is, the better the smoothing effect is, the better the consistency of the abnormal condition of the recognition result is, and the larger the time delay is brought at the same time. The smaller the value of D is, the poorer the smoothing effect is, and the more discontinuous abnormal intervals exist in the recognition result, but the time delay is smaller. Therefore, a recommended value for D is generally around 40 (35-50).
3) Relating to a piece k of historical data
Namely, the historical data reference judgment length, and the parameter is mainly used for calculating the size condition of the historical data and is convenient for calculating the change size condition of the current data and the historical data. Generally, a value of k is preferably about 10.
4) Selection of magnitude of translation value
Because the sizes of data possibly corresponding to different units are obviously different, in order to reduce the interference between the size of the data volume of the data and the threshold judgment, before the data is preprocessed, the data is integrally translated, the size of the translation volume is generally selected according to historical data, and the size of the translation volume is generally selected from the minimum value which can be reached when the corresponding unit normally works.
Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can make modifications and equivalents to the embodiments of the present invention without departing from the spirit and scope of the present invention, which is set forth in the claims of the present application.
Claims (4)
1. An automated power trend identification method, comprising the steps of:
(1) the server receives active power original data in real time;
(2) preprocessing original data;
(3) active power oscillation state identification;
(4) active power trend identification;
in the step (2), a moving average smoothing mode is adopted for preprocessing original data to achieve data smoothing, and the active power variation trend is extracted; the raw data preprocessing comprises the following steps:
1) extracting average values at equal intervals from the raw data, including: inputting original data Dt, extracting an average value of each T points of the original data Dt, recording the average value as 1 point, and storing the 1 point in an A variable; the A variables include A (1), A (2), A (3), A (4.); the A (1), A (2), A (3) and A (4) are Dt (1: T), Dt (T + 1: 2T), Dt (2T + 1: 3T) and Dt (3T + 1: 4T) respectively;
2) moving average processing is carried out on the extracted average value data so as to achieve the purpose of smoothing the data, and the method comprises the following steps: carrying out moving average processing on the variable A by adopting the following formula;
wherein: b is processed data, i represents data subscript parameters, j is cycle parameters of accumulated sum, and D represents translation interval length;
let the length of the data a be Len, and N be the result of Len dividing D by integer, then the value range of the subscript parameter i is 1, 2.
3) Expanding the data after the moving smoothing, comprising: expanding each data point in the data B by 24 values forwards, wherein the expanded value is the value of the current point, and the expanded value is recorded as a variable C after the expansion; the C variables include C (1: T), C (T + 1: 2T), C (2T + 1: 3T), C (3T + 1: 4T.);
the step (3) comprises the following steps:
initializing the following parameters according to different unit characteristics, wherein the parameters comprise:
A. active power oscillation threshold percentage Per; the parameter is used for adjusting the sensitivity of oscillation feature identification, the larger the parameter value is, the smaller the oscillation identification area is, and otherwise, the larger the oscillation identification area is;
B. adding a footmark record amount U _ ind; the system comprises a data processing unit, a data processing unit and a data processing unit, wherein the data processing unit is used for recording the positions of all the subscripts of the maximum extreme points in original data;
C. the lower subscript record amount D _ ind; the system comprises a data processing unit, a data processing unit and a data processing unit, wherein the data processing unit is used for recording the positions of all minimum extreme points in original data;
smoothly preprocessing the original active power data by using the preprocessing steps 1) and 2) in the step (2), and returning the unexpanded preprocessed active power data;
searching upper envelope data Up and lower envelope data Down in the oscillation state;
fourthly, performing linear interpolation on the found upper envelope data Up and lower envelope data Down to obtain interpolated data U _ A and D _ A; the linear interpolation algorithm is as follows:
for the upper envelope data Up, first take the first two data point coordinate information (x1, y1) and (x2, y 2); wherein, the x coordinate represents the time, and the y coordinate represents the active power value of the corresponding time; if x2-x 1> 1, then linear interpolation of the point x between x1 and x2 yields the corresponding y value as follows:
y＝(y2-y1)/(x2-x1)*(x-x1)+y1
then, linear interpolation is carried out on (x2, y2) and (x3, y3), and the U _ A is obtained after the steps are sequentially carried out; in the same way, D _ A is obtained from the lower envelope data Down;
smoothing the interpolated upper envelope data U _ A, interpolated lower envelope data D _ A and original data by respectively reusing the preprocessing step, and marking the finally obtained data as Temp _ U, Temp _ D and M;
sixthly, detecting the oscillation state of the active power according to the oscillation judgment rule of the active power; the active power oscillation judgment rule is as follows: when min ((Temp _ U-M), (M-Temp _ D)) > M × Per, judging that the active power oscillates, and if not, ending;
seventhly, expanding the judgment result to achieve the purpose of matching with the original data quantity;
the step (4) comprises the following steps:
initializing parameters including a historical data segment k, a smooth data interval parameter D, an active power trend threshold percentage Per and a translation amount py according to different unit characteristics;
receiving 25 × D unprocessed active power signal sampling values, performing downward translation and smoothing on active power data by using steps 1) and 2) of a preprocessing method, recording as B, and returning unexpanded data;
<3>according to the active power preprocessing data, carrying out weighted average on the first k data of the current data B (j) to obtain a comprehensive index
Wherein the weight satisfies w (j-k) is less than or equal to w (j- (k-1)). less than or equal to w (j-1), and the threshold value thre ═ C × Per; j is the subscript of the current processing data segment, m is a cycle parameter, w (j-m) is a corresponding weight parameter, and B (k-m +1) represents the value of an array B corresponding to the subscript;
combining the calculation result of the trend judgment comprehensive index, and detecting active power trend judgment by utilizing a trend judgment rule;
<5> executing a trend extension strategy if the active power trend is an ascending trend or a descending trend according to the preliminary judgment result;
and 6, expanding the judgment result, namely expanding each data point in the B data forward by 24, wherein the expanded value is the numerical value of the current point, and the expanded state is the state of the current point.
2. The automatic power trend identification method of claim 1, wherein the active power oscillation threshold percentage Per is between 0.005 and 0.01.
3. The automatic power trend identification method of claim 1, wherein finding the upper envelope of the oscillating state comprises: (a) selecting continuous d data from the starting point of the active power data, searching the maximum data and the corresponding superscript in the d data, selecting the data with the large superscript if two or more maximum data exist, respectively storing the maximum data and the corresponding superscript in arrays Up and U _ ind, and marking the superscript as a variable index; (b) repeating the steps (a) and (b) by taking index +1 as a starting point until all data are inquired and the searching of the upper envelope line in the oscillation state is finished;
finding the lower envelope when oscillating includes: (c) selecting continuous D data from the starting point of the data, finding out the minimum data and the corresponding subscript mark in the D data, if two or more than two minimum data exist, selecting the data with the small subscript mark, respectively storing the minimum data and the corresponding subscript mark in arrays Down and D _ ind, and marking the subscript mark as index'; (d) and (d) repeating the steps (c) and (d) by taking index' +1 as a starting point until all data are inquired and the searching of the lower envelope line in the oscillation state is finished.
4. The automatic power trend identification method of claim 1, wherein the historical data segment k ranges from 8 to 12, and the smooth data interval parameter D ranges from 35 to 50; the value of the translation amount py is selected as the minimum value reached when the corresponding unit normally works.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510673427.2A CN106600449B (en) | 2015-10-16 | 2015-10-16 | Automatic power trend identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510673427.2A CN106600449B (en) | 2015-10-16 | 2015-10-16 | Automatic power trend identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106600449A CN106600449A (en) | 2017-04-26 |
| CN106600449B true CN106600449B (en) | 2021-03-16 |
Family
ID=58553766
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510673427.2A Active CN106600449B (en) | 2015-10-16 | 2015-10-16 | Automatic power trend identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106600449B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110489810B (en) * | 2019-07-24 | 2022-11-25 | 上海金润联汇数字科技有限公司 | Automatic trend extraction method based on data blocks |
| CN113075180A (en) * | 2021-03-24 | 2021-07-06 | 临海市鸥巡电子科技有限公司 | Method and system for detecting change trend of fluorescence data |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002519774A (en) * | 1998-06-23 | 2002-07-02 | トムソン ライセンシング ソシエテ アノニム | Interpolation method using pre-emphasized input samples |
| CN100556014C (en) * | 2006-11-23 | 2009-10-28 | 中兴通讯股份有限公司 | A kind of method of waveforms in signal process identification |
| CN104077309B (en) * | 2013-03-28 | 2018-05-08 | 日电（中国）有限公司 | A kind of method and apparatus that dimension-reduction treatment is carried out to time series |
| CN104485666B (en) * | 2014-12-22 | 2016-11-23 | 大连理工大学 | Total regulation power curve trend recognition methods in a kind of Automatic Generation Control |
-
2015
- 2015-10-16 CN CN201510673427.2A patent/CN106600449B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN106600449A (en) | 2017-04-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103989462B (en) | The extracting method of a kind of pulse wave fisrt feature point and second feature point | |
| CN108814590B (en) | Detection method of electrocardio QRS wave group and electrocardio analysis method thereof | |
| CN108446632B (en) | Partial discharge pulse edge searching and partial discharge confirming method | |
| CN117421687B (en) | Method for monitoring running state of digital power ring main unit | |
| CN116111727B (en) | Comprehensive distribution box abnormity monitoring method based on dynamic temperature threshold | |
| CN105932774A (en) | Device state early warning method in smart transformer substation based on ICA algorithm | |
| CN109255395B (en) | Service life prediction method of ball screw pair | |
| CN104049244A (en) | Radar speed measurement machine carbon brush abrasion identification method based on time domain feature value analysis | |
| CN108814591B (en) | Method for detecting width of electrocardio QRS wave group and electrocardio analysis method thereof | |
| CN111898669A (en) | Machine learning-based direct-current submerged arc furnace abnormal event early warning system | |
| CN106600449B (en) | Automatic power trend identification method | |
| CN116582596B (en) | Data processing method and device and PLC data transmission system based on Internet of things | |
| CN105974223B (en) | A kind of method and system for on-line checking electrical equipment working condition | |
| CN110702986B (en) | Method and system for dynamically generating self-adaptive signal search threshold in real time | |
| CN102608415B (en) | Software frequency tracking algorithm on basis of weighted double fitting | |
| CN116335925A (en) | Data enhancement-based intelligent regulation and control system for underground coal mine emulsification pump station | |
| CN110378624A (en) | Fired power generating unit primary frequency modulation index calculating method and system based on trend abstraction | |
| CN114184870A (en) | Non-invasive load identification method and equipment | |
| CN117150244A (en) | Intelligent power distribution cabinet state monitoring method and system based on electrical parameter analysis | |
| TW201830223A (en) | Time series data processing apparatus and processing method for smoothing processing | |
| CN116784852A (en) | Fatigue state determining method and device, intelligent mattress and medium | |
| CN105183612B (en) | The appraisal procedure of server free memory abnormal growth and operation conditions | |
| CN117172601A (en) | Non-invasive load monitoring method based on residual total convolution neural network | |
| CN115389812B (en) | Artificial neural network short-circuit current zero prediction method and prediction terminal | |
| CN106683001A (en) | Thermal power plant set identification data selection method based on historical operation data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |