CN116031879A - Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system - Google Patents
Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system Download PDFInfo
- Publication number
- CN116031879A CN116031879A CN202310174560.8A CN202310174560A CN116031879A CN 116031879 A CN116031879 A CN 116031879A CN 202310174560 A CN202310174560 A CN 202310174560A CN 116031879 A CN116031879 A CN 116031879A
- Authority
- CN
- China
- Prior art keywords
- feature
- feature selection
- algorithm
- sample
- power system
- 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 33
- 230000001052 transient effect Effects 0.000 title claims abstract description 25
- 238000010187 selection method Methods 0.000 title claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 52
- 238000005457 optimization Methods 0.000 claims abstract description 30
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000012216 screening Methods 0.000 claims abstract description 23
- 238000005259 measurement Methods 0.000 claims abstract description 9
- 238000004088 simulation Methods 0.000 claims abstract description 5
- 238000010276 construction Methods 0.000 claims abstract description 3
- 238000000556 factor analysis Methods 0.000 claims abstract description 3
- 238000004364 calculation method Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 238000013097 stability assessment Methods 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 2
- 230000005284 excitation Effects 0.000 claims description 2
- 238000002372 labelling Methods 0.000 claims 1
- 238000013210 evaluation model Methods 0.000 abstract description 6
- 230000008569 process Effects 0.000 abstract description 5
- 230000000694 effects Effects 0.000 description 7
- 230000008901 benefit Effects 0.000 description 5
- 238000001914 filtration Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000004806 packaging method and process Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000007418 data mining Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 241000254032 Acrididae Species 0.000 description 1
- 241000238814 Orthoptera Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003795 chemical substances by application Substances 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000019637 foraging behavior Effects 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Classifications
-
- 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
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a hybrid intelligent feature selection method suitable for transient voltage stability evaluation of a power system, which mainly comprises the following steps: s1, sample generation: the method comprises the steps of obtaining a high-dimensional time sequence sample data set through stability related factor analysis, original feature set construction and multi-scene transient time domain simulation; s2, feature effectiveness measurement and preliminary screening based on a T-Relief algorithm; s3, feature selection and stability evaluation based on an improved population intelligent algorithm: the search performance of the group intelligent optimization algorithm is enhanced through the feature validity metric value obtained in the step S2, and an improved group intelligent optimization algorithm is obtained; based on the algorithm, a packaged feature selection scheme is built, a ConvGRU evaluation model is embedded as a subset evaluator, feature subset optimization is further achieved on the basis of primary screening of features in step S2, and feature redundancy is reduced. The method can directly process the time sequence characteristics, and can improve the searching efficiency as much as possible while guaranteeing the screening quality of the characteristic subsets.
Description
Technical Field
The invention relates to the technical field of power systems, in particular to a hybrid intelligent feature selection method suitable for transient voltage stability evaluation of a power system.
Background
The new energy grid connection and the operation of multi-circuit extra-high voltage direct current transmission engineering in large scale make the characteristics of the novel power system of double high and low and the load center of hollow obvious and the uncertainty of source-load aggravated. The high power electronization causes the dynamic characteristics of the power grid to change deeply, meanwhile, the fault tolerance capability is reduced, and the risk of transient voltage instability of the system is aggravated. With the development and maturity of big data theory and artificial intelligence technology and the rapid popularization of power grid measuring devices, the response data-driven artificial intelligence method provides a new thought for rapidly realizing transient voltage stability evaluation. However, the actual power system has a large number of various elements, the power grid is large in scale, the electrical characteristics are high-dimensional data, and redundancy among the characteristics can greatly influence classification performance and efficiency of the evaluation model. In recent years, in order to fully mine mechanism evolution information of key features of a disturbed system so as to improve stability evaluation accuracy, a part of literature proposes to use dynamic time sequence data for transient stability analysis, which undoubtedly aggravates calculation burden and overfitting risk of a model. Thus, exploring suitable feature selection schemes to reduce the original feature dimensions is a key issue for application of artificial intelligence methods in power system transient voltage stability assessment.
In the prior researches, key feature selection is carried out manually, namely, variable features conforming to priori knowledge are selected by means of expert experience to carry out stability analysis; many students currently conduct feature selection research based on data mining, and there are three types of filtering, embedding and packaging. The filtering method is used for screening the features most relevant to the target attribute by calculating the association degree between the feature variable and the label, and typical association measurement criteria comprise a Relief statistic, an information measure, a Fisher statistic and the like. The embedded method synchronously realizes feature selection in the training process of the learning model, and a typical embedded feature selection model is provided with a decision tree. The packaging method is to find out the best feature subset matched with the learning model in an iterative optimization mode through the mutual matching of the optimization algorithm and the learning model.
Because the fault characteristics of a plurality of novel dynamic response devices in the current power system are unknown, the mechanism analysis of transient voltage instability is also unknown, the variable characteristics are difficult to adapt to the characteristics of a complex large power grid and information omission is easy to cause only by relying on manual experience. In the feature selection research based on data mining, although the filtering method has high calculation speed and strong operability, the feature screening principle is single, the feature redundancy cannot be effectively reduced, and the filtering method cannot be effectively applied to time sequence feature selection; the embedded method is too dependent on a learning model, so that the embedded method is difficult to adapt to the requirement of transient voltage stability rapid evaluation; the accuracy rate of the feature evaluation selected by the packaging method is higher, but the algorithm calculation cost is high, and the calculation efficiency is required to be improved. Therefore, the existing feature selection method has the defect of considering the screening efficiency and the screening accuracy.
Disclosure of Invention
Aiming at the problem that the screening efficiency and the classification performance of the feature subset are difficult to achieve in the current power system analysis feature selection method, the invention provides the hybrid intelligent feature selection method suitable for transient voltage stability evaluation of the power system.
The invention provides a hybrid intelligent feature selection method suitable for transient voltage stability evaluation of a power system, which comprises the following steps:
s1, sample generation:
and obtaining a high-dimensional time sequence sample data set by carrying out stability related factor analysis, original feature set construction and multi-scene transient time domain simulation. And obtaining transient voltage sample data of the power system after being disturbed by the time domain simulation, and completing sample stability marking according to engineering practical criteria, thereby constructing training and testing data sets of the feature selection and evaluation model, namely a high-dimensional time sequence sample data set.
S2, feature effectiveness measurement and preliminary screening based on a T-Relief algorithm.
The Relief algorithm has the advantage of high operation efficiency, but is not directly applicable to time sequence input feature selection. It is improved by a time-sequential layering process, resulting in a T-Relief algorithm. The algorithm is used for preliminary screening of original features, on one hand, feature classification effectiveness measurement values are calculated to enhance the searching performance of the subsequent step S3, on the other hand, preliminary dimension reduction is realized, and the iterative operation cost of the step S3 is reduced. The method specifically comprises the following substeps:
s21, assuming that an original characteristic data set has M samples, wherein each sample has d characteristic attributes, and the number of time points of data record is N; performing time sequence layering on M time sequence high-dimensional characteristic data with N time steps to form N layered high-dimensional characteristic matrixes;
s22, respectively calculating Euclidean distances between M samples and other samples in the N layered high-dimensional feature matrices, and constructing an Euclidean distance matrix;
s23, searching for neighbor samples of each layered high-dimensional characteristic sample according to the Euclidean distance matrix; calculating the relevant statistics of each feature at the corresponding moment;
s24, averaging the related statistics obtained under different time layering to obtain each characteristic effectiveness metric delta of the comprehensive time sequence information;
s25, setting a threshold tau, screening out effective features meeting the conditions, and carrying out the next step. The set threshold τ may be the number of screening features or a specific value of the relevant statistic, etc.
S3, feature selection and stability evaluation based on an improved population intelligent algorithm.
In the step, the search performance of the group intelligent optimization algorithm (BGOA) is enhanced through the feature validity metric value obtained in the step S2, and the improved group intelligent optimization algorithm (IBGOA) is obtained. Based on the algorithm, a packaged feature selection scheme is built, a ConvGRU evaluation model is embedded as a subset evaluator, time sequence evolution information and an internal combination relation of features can be fully considered, feature subset optimization is further achieved on the basis of primary screening of features in step S2, and feature redundancy is reduced.
The method comprises the following substeps:
s31, adopting an improved binary locust optimization algorithm to perform feature selection;
the binary locust optimization algorithm is subjected to search performance enhancement through the effectiveness metric delta obtained in the step S24, so that an improved binary locust optimization algorithm is obtained; in the improved binary locust optimization algorithm, an improved locust position initialization formula and an iterative update formula are as follows:
wherein a epsilon (0, 1) is a weight coefficient, and r epsilon [0,1] is a uniformly distributed random number; round () represents a rounding function;
wherein, beta, eta and gamma are weight coefficients, and r E [0,1] is a uniformly distributed random number;
s32, classifying performance evaluation is carried out on the feature subsets by a multi-dimensional time sequence classification model embedded with ConvGRU, whether the comprehensive evaluation index meets the iteration number requirement is judged by taking the fitness function as the comprehensive evaluation index, if the comprehensive evaluation index meets the iteration number requirement, the optimal feature subsets are output, and if the iteration number requirement is not met, the feature selection is carried out by repeatedly adopting an improved binary locust optimization algorithm again.
In this step, the ConvGRU calculation formula is as follows:
R t ＝σ(W xr *X t +W hr *H t-1 )
Z t ＝σ(W xz *X t +W hz *H t-1 )
wherein σ and tanh represent excitation functions; the addition of the elements is carried out; * Is convolution operation; r is R t A schematic reset gate; z is Z t Schematic update gate; x is X t Representing the current time input data;
representing the current candidate value; h t-1 Representing the hidden state at the previous moment; w (W) xr 、W hr 、W xz 、W hz 、W xh 、W hh Respectively representing the weight coefficients of the corresponding connections.
The fitness function employed is as follows:
wherein, beta epsilon (0, 1) is a weight coefficient, r| is the dimension of the corresponding feature subset, and N| is the original feature dimension; p (P) c Is a comprehensive erroneous judgment rate index;
comprehensive erroneous judgment rate index P c The calculation formula is as follows:
P c ＝αP fs +(1-α)P fus
wherein, the miss rate
Misjudgment rate
Wherein alpha is a weight coefficient, F s Indicating the number of samples that were erroneously determined to be stable (missed determination); f (F) us The number of samples (erroneous judgment) indicating erroneous judgment as unsteady; t (T) s Indicating the number of samples correctly determined to be stable; t (T) us The number of samples correctly determined as unstable is represented.
Compared with the prior art, the invention has the following advantages:
(1) The time sequence of the Relief algorithm is improved to obtain a T-Relief algorithm, and the method is suitable for the effectiveness measurement of high-dimensional time sequence characteristics and still has the advantage of high operation efficiency;
(2) The feature validity metric value is fused, so that the initialization and iterative updating formulas of the group intelligent optimization algorithm are improved, and the efficiency and performance of the package type feature selection based on the intelligent algorithm are effectively improved;
(3) The mixed intelligent feature selection method based on double screening can effectively realize feature dimension reduction. The primary screening stage provides characteristic effectiveness metric values and primarily reduces the dimension, so that the efficiency of the subsequent subset optimization is improved. ConvGRU time sequence evaluation models are embedded in a packaging stage based on an improved group intelligent optimization algorithm, so that time-varying characteristics of the features can be fully considered, and higher classification precision is achieved. Compared with other common feature selection methods, the feature subset classification performance selected by the method is better.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Drawings
FIG. 1, a block diagram of a hybrid intelligent feature selection algorithm.
FIG. 2 is a flow chart of the improved T-Relief algorithm of the present invention.
FIG. 3, convGRU unit structure schematic diagram.
Fig. 4, IBGOA and BGOA iteration convergence curves.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
As shown in FIG. 1, the hybrid intelligent feature selection method for adapting to transient voltage stability evaluation of a power system mainly comprises three parts, namely sample generation, feature effectiveness measurement and preliminary screening based on a T-Relief algorithm, feature subset optimizing based on a group intelligent optimization algorithm and stability evaluation.
The principle of the T-Relief algorithm is as follows:
the Relief algorithm is a feature selection method based on a filtering formula and aims at the two-classification problem, and the method calculates relevant statistics according to the distinguishing capability of features on a close-range sample, so that effectiveness evaluation is carried out on different features. The method comprises the following specific steps:
(1) Based on the Euclidean distance, searching the sample closest to the appointed sample in the similar samples as a 'guess neighbor', searching the sample closest to the appointed sample in the non-similar samples as a 'guess neighbor', and then calculating the relevant statistics of each feature according to the selected 'neighbor in guess' and 'neighbor in guess', and finally summing the relevant statistics of all samples in the training set to obtain the overall feature importance measurement result, wherein the calculation formula is as follows:
wherein the superscript j represents the j-th dimension feature of the corresponding sample, delta is the relevant statistic list of all features, and x i For the current calculation sample, x i,nh "neighbor in guess" for current sample, x i,nm The "false neighbor" for the current sample. Because the power system response data is a continuous variable, the power system response data is a continuous variable
The values of (2) are as follows:
(2) And sequencing the obtained feature related statistics, selecting a proper threshold value, and selecting the features with higher importance to perform classification tasks.
In the method, the time sequence of the Relief algorithm is improved through time sequence layering processing, and the T-Relief algorithm obtained after the improvement is obtained, so that the T-Relief algorithm can be directly used for screening high-dimensional time sequence characteristics. Assuming that the original feature data set has M samples, each sample has d feature attributes, and the number of time points of the data record is N, the calculation flow of the T-Relief algorithm is shown in fig. 2, and specifically is as follows:
(1) performing time sequence layering on M time sequence high-dimensional characteristic data with N time steps to form N layered high-dimensional characteristic matrixes;
(2) respectively calculating Euclidean distances between M samples and other samples in the N layered high-dimensional feature matrices, and constructing an Euclidean distance matrix;
(3) searching for a neighbor sample of each layered high-dimensional characteristic sample according to the Euclidean distance matrix, normalizing the original characteristic data, and calculating the relevant statistics of each characteristic at the corresponding moment; the calculation formula is as follows:
(4) averaging the related statistics obtained under different time layering to obtain each characteristic effectiveness metric value of the comprehensive time sequence information;
(5) and setting a threshold tau, screening out effective characteristics meeting the conditions, and carrying out next screening research. The threshold τ in the present method may be the number of screening features or a particular value of the relevant statistic.
The improved intelligent optimization algorithm principle is as follows:
the group intelligent-based packaged feature selection method is an important means for searching an optimal feature subset approximate solution at present. The invention selects a relatively stable and reliable binary locust optimization algorithm (Binary grasshopper optimization algorithm, BGOA) as a search strategy for the selection of the package type features.
BGOA is a group intelligent optimization algorithm for carrying out mathematical modeling according to social acting force among individuals, which is provided by simulating locust foraging behaviors, and has the advantages of simple algorithm structure and stable performance, wherein the mathematical model is as follows:
X i ＝S i +G i +A i
wherein, the subscript i represents the ith locust in the population; x is X i Representing its position; s is S i Social forces to other agents; g i Is self-gravity; a is that i For the wind power, the gravity and wind power effect are usually ignored in practical application. Wherein S is i The calculation formula of (2) is as follows:
wherein d ij ＝|x i -x j The i represents the distance between two locusts, s () is a social strength function, the positive number represents attraction, the negative number represents repulsion, and the calculation formula is as follows:
wherein f is an gravitational strength parameter, l is an gravitational range parameter, f=0.5, l=1.5, r is an independent variable, and is not particularly defined, and is equivalent to d in the above formula ij Is the code of (2).
In the feature selection problem, BGOA position vector X i For binary sequences of length D, representing the selection of D features, i.e. X i d=1 means that the d-th dimension feature is selected and vice versa. Therefore, combining with the social force effect, the stepping vector of locust position update is defined as Deltax i And the S-shaped transfer function is utilized to convert the probability into the selected probability, and the specific calculation mode is as follows:
wherein u is d ，l d The upper limit and the lower limit of the characteristic value are respectively; t represents an iteration round; t () is a function; c is an adaptive adjustment coefficient, and the adjustment mode is as follows:
c＝c max -t(c max -c min )/L
wherein, c max ，c min The maximum adjustment coefficient and the minimum adjustment coefficient are respectively, and L represents the maximum iteration number.
In the invention, for fusing the guiding effect of characteristic validity metric values at the T-Relief stage on the locust iterative update, an improved locust position initialization formula and an iterative update formula are as follows:
wherein a epsilon (0, 1) is a weight coefficient, and r epsilon [0,1] is a uniformly distributed random number. Round () represents a rounding function,
wherein, beta, eta and gamma are weight coefficients, and r E [0,1] is a uniformly distributed random number.
The convglu unit structure is shown in fig. 3. The evaluation of the quality of the feature subset to be selected is mainly considered from two aspects, namely, the classification effect of using the feature subset for stable evaluation and the dimension of the feature subset.
To evaluate classifier performance, define an accuracy rate P acc Rate of missed judgment P fs Erroneous judgment rate P fus The metrics evaluate the performance of the transient classifier (convglu). Simultaneously introducing weight coefficient alpha>1 to balance the importance degree of misjudgment and missed judgment, and define the index P of the comprehensive misjudgment rate c To evaluate the classification performance of the feature subset.
P c ＝αP fs +(1-α)P fus
To achieve maximization of classification accuracy and minimization of feature subset dimensions, an objective function for stable evaluation feature selection (i.e., an optimization fitness function that improves the BGOA feature selection algorithm) is defined as:
wherein, beta epsilon (0, 1) is a weight coefficient, r| is the dimension of the corresponding feature subset, and N| is the original feature dimension.
Performance comparative analysis:
(1) Improved BGOA algorithm (IBGOA) performance comparison with original BGOA algorithm
In order to verify that the improved BGOA algorithm (IBGOA) has better performance in the subset iterative optimization process, a comparison experiment is carried out on the improved BGOA algorithm (IBGOA) and the subset iterative optimization process. Experiment setting: and (3) respectively carrying out feature screening on the original feature set by using BGOA and IBGOA, wherein the iteration times are 50, and the fitness average value of each iteration is obtained through multiple tests. The final fitness value iteration convergence curve is shown in fig. 4. The figure shows that the optimized searching efficiency of the subset is higher and the adaptability value of the optimal subset is lower under the IBGOA method, so that the effectiveness of the invention in improving BGOA is fully proved.
(2) Transient voltage stability evaluation effect comparison before and after feature selection
And selecting a stable evaluation accuracy rate, a missed judgment rate, a misjudgment rate index, a characteristic dimension and a model training time angle to carry out comparison and demonstration on the effect of characteristic selection.
Experimental details: after the original feature (370 dimensions) is calculated by the T-Relief, the 131 dimensions of the effective feature with the effectiveness measurement value larger than 0.6 are reserved; and after IBGOA iterative optimization, the key characteristics are finally kept in 18 dimensions (average value is obtained through multiple tests, and each test iterates 50 times), and the experimental results are shown in table 1. As can be seen from Table 1, after the feature selection is performed by the method of the present invention, only 18 key features are finally retained, and the model training time is even reduced by 76.41%. In the case of a deepened model complexity and a sharply increased sample size, this will greatly reduce the difficulty of model training. In addition, although the dimension of the feature is greatly reduced, the accuracy of model evaluation is increased by 2.73%, and the misjudgment rate and the missed judgment rate are both reduced by 1.365%, which indicates that the feature selection method provided by the invention can effectively remove invalid features in high-dimension features, thereby improving the accuracy of the transient voltage stability evaluation model.
TABLE 1 comparison of stability evaluation effects before and after feature selection
The present invention is not limited to the above-mentioned embodiments, but is intended to be limited to the following embodiments, and any modifications, equivalents and modifications can be made to the above-mentioned embodiments without departing from the scope of the invention.
Claims (4)
1. A hybrid intelligent feature selection method suitable for transient voltage stability evaluation of a power system is characterized by comprising the following steps:
s1, sample generation:
the method comprises the steps of obtaining a high-dimensional time sequence sample data set through stability related factor analysis, original feature set construction and multi-scene transient time domain simulation;
s2, feature effectiveness measurement and preliminary screening based on a T-Relief algorithm, comprising the following substeps:
s21, assuming that an original characteristic data set has M samples, wherein each sample has d characteristic attributes, and the number of time points of data record is N; performing time sequence layering on M time sequence high-dimensional characteristic data with N time steps to form N layered high-dimensional characteristic matrixes;
s22, respectively calculating Euclidean distances between M samples and other samples in the N layered high-dimensional feature matrices, and constructing an Euclidean distance matrix;
s23, searching for neighbor samples of each layered high-dimensional characteristic sample according to the Euclidean distance matrix; calculating the relevant statistics of each feature at the corresponding moment;
s24, averaging the related statistics obtained under different time layering to obtain each characteristic effectiveness metric delta of the comprehensive time sequence information;
s25, setting a threshold tau, screening out effective features meeting the conditions, and carrying out the next step;
s3, feature selection and stability evaluation based on an improved population intelligent algorithm, comprising the following substeps:
s31, adopting an improved binary locust optimization algorithm to perform feature selection;
the binary locust optimization algorithm is subjected to search performance enhancement through the effectiveness metric delta obtained in the step S24, so that an improved binary locust optimization algorithm is obtained; in the improved binary locust optimization algorithm, an improved locust position initialization formula and an iterative update formula are as follows:
wherein a epsilon (0, 1) is a weight coefficient, and r epsilon [0,1] is a uniformly distributed random number; round () represents a rounding function;
wherein, beta, eta and gamma are weight coefficients, and r E [0,1] is a uniformly distributed random number;
s32, classifying performance evaluation is carried out on the feature subsets by a multi-dimensional time sequence classification model embedded with ConvGRU, an fitness function is used as a comprehensive evaluation index, whether the comprehensive evaluation index meets the iteration number requirement is judged, if the comprehensive evaluation index meets the iteration number requirement, the optimal feature subset is output, and if the comprehensive evaluation index does not meet the iteration number requirement, an improved binary locust optimization algorithm is repeatedly adopted again to carry out feature selection.
2. The hybrid intelligent feature selection method for adapting to power system transient voltage stability assessment according to claim 1, wherein in step S1, transient voltage sample data after power system disturbance is obtained through time domain simulation, and sample stability labeling is completed according to engineering practical criteria, so as to construct training and testing data sets of feature selection and assessment models, namely high-dimensional time sequence sample data sets.
3. The hybrid intelligent feature selection method for adapting to power system transient voltage stability assessment according to claim 1, wherein in step S32, the calculation formula of convglu is as follows:
R t ＝σ(W xr *X t +W hr *H t-1 )
Z t ＝σ(W xz *X t +W hz *H t-1 )
wherein σ and tanh represent excitation functions; the addition of the elements is carried out; * Is convolution operation; r is R t Schematic weightSetting a door; z is Z t Schematic update gate; x is X t Representing the current time input data;
representing the current candidate value; h t-1 Representing the hidden state at the previous moment; w (W) xr 、W hr 、W xz 、W hz 、W xh 、W hh Respectively representing the weight coefficients of the corresponding connections.
4. The hybrid intelligent feature selection method for adapting to transient voltage stability assessment of a power system according to claim 3, wherein in step S32, the fitness function employed is as follows:
wherein, beta epsilon (0, 1) is a weight coefficient, r| is the dimension of the corresponding feature subset, and N| is the original feature dimension; p (P) c Is a comprehensive erroneous judgment rate index;
P c the calculation formula is as follows:
P c ＝αP fs +(1-α)P fus
wherein, alpha is a weight coefficient,
miss rate P fs The calculation formula of (2) is as follows:
false positive rate P fus The calculation formula of (2) is as follows:
wherein F is s Indicating the number of samples that were erroneously determined to be stable; f (F) us A sample number indicating that the error determination is unstable; t (T) s Representation ofCorrectly judging the stable sample number; t (T) us The number of samples correctly determined as unstable is represented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310174560.8A CN116031879A (en) | 2023-02-28 | 2023-02-28 | Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310174560.8A CN116031879A (en) | 2023-02-28 | 2023-02-28 | Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116031879A true CN116031879A (en) | 2023-04-28 |
Family
ID=86076116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310174560.8A Pending CN116031879A (en) | 2023-02-28 | 2023-02-28 | Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116031879A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574116A (en) * | 2024-01-15 | 2024-02-20 | 湖南大学 | Mixed feature selection method for transient stability evaluation of seabed direct current system |
| CN117909333A (en) * | 2024-02-02 | 2024-04-19 | 深圳天朴科技有限公司 | Screening method and system for realizing data based on big data combined with artificial intelligence |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102324743A (en) * | 2011-09-21 | 2012-01-18 | 国网电力科学研究院 | The online transient safe and stable assessment of electric power system forecast failure screening technique |
| CN111478314A (en) * | 2020-03-20 | 2020-07-31 | 广西电网有限责任公司电力科学研究院 | Transient stability assessment method for power system |
| CN111523778A (en) * | 2020-04-10 | 2020-08-11 | 三峡大学 | Power grid operation safety assessment method based on particle swarm algorithm and gradient lifting tree |
| CN112632840A (en) * | 2020-12-04 | 2021-04-09 | 山东大学 | Power grid transient stability evaluation method based on adaptive differential evolution algorithm and ELM |
| CN113054653A (en) * | 2019-12-27 | 2021-06-29 | 南京理工大学 | Power system transient stability evaluation method based on VGGNet-SVM |
| CN113468817A (en) * | 2021-07-13 | 2021-10-01 | 淮阴工学院 | Ultra-short-term wind power prediction method based on IGOA (optimized El-electric field model) |
| CN114881101A (en) * | 2022-03-21 | 2022-08-09 | 武汉大学 | Power system typical scene associated feature selection method based on bionic search |
-
2023
- 2023-02-28 CN CN202310174560.8A patent/CN116031879A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102324743A (en) * | 2011-09-21 | 2012-01-18 | 国网电力科学研究院 | The online transient safe and stable assessment of electric power system forecast failure screening technique |
| CN113054653A (en) * | 2019-12-27 | 2021-06-29 | 南京理工大学 | Power system transient stability evaluation method based on VGGNet-SVM |
| CN111478314A (en) * | 2020-03-20 | 2020-07-31 | 广西电网有限责任公司电力科学研究院 | Transient stability assessment method for power system |
| CN111523778A (en) * | 2020-04-10 | 2020-08-11 | 三峡大学 | Power grid operation safety assessment method based on particle swarm algorithm and gradient lifting tree |
| CN112632840A (en) * | 2020-12-04 | 2021-04-09 | 山东大学 | Power grid transient stability evaluation method based on adaptive differential evolution algorithm and ELM |
| CN113468817A (en) * | 2021-07-13 | 2021-10-01 | 淮阴工学院 | Ultra-short-term wind power prediction method based on IGOA (optimized El-electric field model) |
| CN114881101A (en) * | 2022-03-21 | 2022-08-09 | 武汉大学 | Power system typical scene associated feature selection method based on bionic search |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117574116A (en) * | 2024-01-15 | 2024-02-20 | 湖南大学 | Mixed feature selection method for transient stability evaluation of seabed direct current system |
| CN117574116B (en) * | 2024-01-15 | 2024-04-19 | 湖南大学 | Mixed feature selection method for transient stability evaluation of seabed direct current system |
| CN117909333A (en) * | 2024-02-02 | 2024-04-19 | 深圳天朴科技有限公司 | Screening method and system for realizing data based on big data combined with artificial intelligence |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116031879A (en) | Hybrid intelligent feature selection method suitable for transient voltage stability evaluation of power system | |
| CN105488528B (en) | Neural network image classification method based on improving expert inquiry method | |
| CN111160474A (en) | Image identification method based on deep course learning | |
| CN111860658A (en) | Transformer fault diagnosis method based on cost sensitivity and integrated learning | |
| CN112232511B (en) | Automatic compression method and platform for pre-training language model for multiple tasks | |
| CN112685504B (en) | Production process-oriented distributed migration chart learning method | |
| CN112039687A (en) | Small sample feature-oriented fault diagnosis method based on improved generation countermeasure network | |
| CN112560079B (en) | Hidden false data injection attack method based on deep belief network and migration learning | |
| CN111368920A (en) | Quantum twin neural network-based binary classification method and face recognition method thereof | |
| CN114019370B (en) | Motor fault detection method based on gray level image and lightweight CNN-SVM model | |
| CN110287985B (en) | Depth neural network image identification method based on variable topology structure with variation particle swarm optimization | |
| CN113139594B (en) | Self-adaptive detection method for airborne image unmanned aerial vehicle target | |
| CN112578089B (en) | Air pollutant concentration prediction method based on improved TCN | |
| CN113177587B (en) | Generalized zero sample target classification method based on active learning and variational self-encoder | |
| CN109214444B (en) | Game anti-addiction determination system and method based on twin neural network and GMM | |
| CN113361685A (en) | Knowledge tracking method and system based on learner knowledge state evolution expression | |
| CN112270334B (en) | Few-sample image classification method and system based on abnormal point exposure | |
| CN118277770A (en) | Obstacle sensing method and device, electronic equipment and storage medium | |
| CN112560948A (en) | Eye fundus map classification method and imaging method under data deviation | |
| CN116090389A (en) | Electric Lu Liang rate estimation method and prediction system based on failure edge sampling | |
| CN115879509A (en) | Convolutional neural network structure optimization method based on agent-assisted evolution algorithm | |
| CN113379037B (en) | Partial multi-mark learning method based on complementary mark cooperative training | |
| CN107590538A (en) | A kind of dangerous source discrimination based on online Sequence Learning machine | |
| CN118095570A (en) | Intelligent load prediction method and system for transformer area, electronic equipment, medium and chip | |
| CN114387474A (en) | Small sample image classification method based on Gaussian prototype classifier |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20230428 |