CN-119518753-B - Method and device for scheduling discrete action equipment of power distribution network and nonvolatile storage medium

CN119518753BCN 119518753 BCN119518753 BCN 119518753BCN-119518753-B

Abstract

The application discloses a power distribution network discrete action equipment scheduling method, a power distribution network discrete action equipment scheduling device and a nonvolatile storage medium. The method comprises the steps of determining load prediction data of a power distribution network and active output prediction data of a distributed power supply of the power distribution network within a preset time period, calling a discrete action equipment scheduling model to process the load prediction data and the active output prediction data so as to obtain a scheduling scheme of discrete action equipment in the power distribution network, wherein training data of the discrete action equipment scheduling model comprise tag data and non-tag data, the discrete action equipment scheduling model comprises a feature extractor used for identifying data input into the discrete action equipment scheduling model, and adjusting tap switching positions of the discrete action equipment within the preset time period according to the scheduling scheme. The method solves the technical problem that the discrete action equipment in the power distribution network cannot be automatically scheduled due to inaccurate output results of the discrete action equipment scheduling model in the related technology.

Inventors

YANG BO
WANG MAOZE
GU Yu
Hou Zongxiang
ZHANG CIHANG
SONG JIAJU
WANG HAIFENG
LI YANG
WU YUTONG
LIU CHUAN
FU ZHE
LI HONGCHUAN
WANG YUNPENG
MA YILAN
SONG XINYU

Assignees

国网北京市电力公司
国家电网有限公司

Dates

Publication Date: 20260505
Application Date: 20241119

Claims (10)

1. The power distribution network discrete action equipment scheduling method is characterized by comprising the following steps of: Load prediction data of a power distribution network in a preset time period and active output prediction data of a distributed power supply of the power distribution network are determined; Invoking a discrete action equipment scheduling model to process the load prediction data and the active output prediction data so as to obtain a scheduling scheme of the discrete action equipment in the power distribution network, wherein the scheduling scheme comprises a predicted switching position of a tap of the discrete action equipment, training data of the discrete action equipment scheduling model comprises tag data and unlabeled data, the discrete action equipment scheduling model comprises a feature extractor for identifying data input into the discrete action equipment scheduling model, and an estimation module for generating the scheduling scheme; Adjusting the tap switching position of the discrete action equipment in the preset time period according to the scheduling scheme; The discrete action device scheduling model is trained by: Acquiring historical data of the power distribution network, wherein the historical data comprises historical load prediction data and historical active output prediction data; determining tagged historical data and untagged historical data in the historical data; determining main component characteristics of the tagged historical data, and screening the tagged historical data according to the main component characteristics to obtain a first training data set; taking the principal component characteristics of the labeled historical data as reference principal component characteristics of the unlabeled historical data, and carrying out clustering screening on the unlabeled historical data according to the reference principal component characteristics to obtain a second training data set; training the discrete action device scheduling model according to the first training data set and the second training data set by adopting an countermeasure generation network.
2. The power distribution network discrete action equipment scheduling method according to claim 1, wherein invoking a discrete action equipment scheduling model to process the load prediction data and the active output prediction data, thereby obtaining a scheduling scheme of discrete action equipment in the power distribution network comprises: Extracting data features from the load prediction data and the active output prediction data through a feature extractor in the discrete action equipment scheduling model, and performing data dimension reduction processing and global average pooling processing on the data features to obtain data feature vectors with preset lengths; And determining a position predicted value of the data feature vector map through a full connection layer by an estimation module in the equipment scheduling model, wherein the position predicted value is used for indicating the predicted switching position.
3. The power distribution network discrete action device scheduling method of claim 1, wherein determining the principal component characteristics of the tagged historical data comprises: Carrying out standardization processing on the tagged data so that the characteristic mean value of the tagged data is a first preset value and the variance is a second preset value; determining a covariance matrix of the standardized tagged data, wherein the principal component features comprise principal component feature vectors, and the covariance matrix is used for reflecting the interrelation between the feature quantities in the tagged data; performing eigenvalue decomposition processing on the covariance matrix to obtain eigenvalues and eigenvectors corresponding to the eigenvalues; and determining the principal component feature vector in the tagged data according to the magnitude of the feature value.
4. The method for scheduling discrete action equipment of a power distribution network according to claim 1, wherein screening the tagged historical data according to the principal component characteristics to obtain a first training data set comprises: Determining a correlation between the principal component characteristics and node voltage characteristics of the power distribution network; and screening the tagged data according to the correlation to obtain the first training data set.
5. The method for scheduling discrete action equipment in a power distribution network according to claim 4, wherein clustering the unlabeled historical data according to the reference principal component features to obtain a second training data set comprises: carrying out standardization processing on the non-tag data, so that the characteristic mean value of the non-tag data is a first preset value, and the variance is a second preset value; constructing a feature space according to the interrelationship reflected by the covariance matrix and the label-free data after the standardization processing; determining a principal component space according to the reference principal component, and processing the feature space in the principal component space to obtain a feature representation corresponding to the unlabeled historical data; And performing cluster analysis processing on the characteristic representation to obtain the second training data set, and verifying the influence mode and influence degree of the tap position of the discrete action equipment in the label-free data on the voltage control result in the process of cluster analysis processing.
6. The power distribution network discrete action device scheduling method as claimed in claim 1, wherein said method further comprises: in the training process, determining cross entropy loss of the discrete action equipment scheduling model, determining a loss function value of the discrete action equipment scheduling model according to the cross entropy loss, and a gear switching penalty coefficient of discrete action equipment in the power distribution network, wherein the gear switching penalty coefficient is used for adjusting the loss function value, and the larger the gear switching penalty coefficient is, the larger the adjusted loss function value is.
7. A power distribution network discrete action device scheduling apparatus, characterized by comprising: the first processing module is used for determining load prediction data of the power distribution network in a preset time period and active output prediction data of a distributed power supply of the power distribution network; The second processing module is used for calling a discrete action equipment scheduling model to process the load prediction data and the active output prediction data so as to obtain a scheduling scheme of the discrete action equipment in the power distribution network, wherein the scheduling scheme comprises a predicted switching position of a tap of the discrete action equipment, training data of the discrete action equipment scheduling model comprises tag data and non-tag data, the discrete action equipment scheduling model comprises a feature extractor used for identifying data input into the discrete action equipment scheduling model, and an estimation module used for generating the scheduling scheme; the third processing module is used for adjusting the tap switching position of the discrete action equipment in the preset time period according to the scheduling scheme; The discrete action device scheduling model is trained by: Acquiring historical data of the power distribution network, wherein the historical data comprises historical load prediction data and historical active output prediction data; determining tagged historical data and untagged historical data in the historical data; determining main component characteristics of the tagged historical data, and screening the tagged historical data according to the main component characteristics to obtain a first training data set; taking the principal component characteristics of the labeled historical data as reference principal component characteristics of the unlabeled historical data, and carrying out clustering screening on the unlabeled historical data according to the reference principal component characteristics to obtain a second training data set; training the discrete action device scheduling model according to the first training data set and the second training data set by adopting an countermeasure generation network.
8. A non-volatile storage medium, wherein a program is stored in the non-volatile storage medium, and wherein the program, when executed, controls a device in which the non-volatile storage medium is located to perform the power distribution network discrete action device scheduling method according to any one of claims 1 to 6.
9. An electronic device comprising a memory and a processor for executing a program stored in the memory, wherein the program is executed to perform the power distribution network discrete action device scheduling method of any one of claims 1 to 6.
10. A computer program product comprising a computer program which, when executed by a processor, implements the power distribution network discrete action device scheduling method according to any one of claims 1 to 6.

Description

Method and device for scheduling discrete action equipment of power distribution network and nonvolatile storage medium Technical Field The application relates to the field of electrical engineering, in particular to a power distribution network discrete action equipment scheduling method and device and a nonvolatile storage medium. Background In order to ensure the accuracy of the model scheduling result, the discrete action equipment scheduling model in the related technology needs to rely on a large amount of reliable labeled data to train the model in the training stage of the model, and the model cannot be trained by adopting unlabeled data. However, a large amount of unlabeled data often exists in the data in the power distribution network, so that the accuracy of the output result of the discrete action equipment scheduling model cannot be ensured in the related technology, and the discrete action equipment in the power distribution network cannot be automatically scheduled. In view of the above problems, no effective solution has been proposed at present. Disclosure of Invention The embodiment of the application provides a method and a device for dispatching discrete action equipment of a power distribution network and a nonvolatile storage medium, which at least solve the technical problem that the discrete action equipment in the power distribution network cannot be automatically dispatched due to inaccurate output result of a discrete action equipment dispatching model in the related technology. According to one aspect of the embodiment of the application, a power distribution network discrete action equipment scheduling method is provided, and comprises the steps of determining load prediction data of a power distribution network and active power output prediction data of a distributed power supply of the power distribution network in a preset time period, calling a discrete action equipment scheduling model to process the load prediction data and the active power output prediction data so as to obtain a scheduling scheme of discrete action equipment in the power distribution network, wherein the scheduling scheme comprises a predicted switching position of a tap of the discrete action equipment in the preset time period, training data of the discrete action equipment scheduling model comprises label data and label-free data, a feature extractor used for identifying the data input into the discrete action equipment scheduling model is included in the discrete action equipment scheduling model, and an estimation module used for generating the scheduling scheme, and adjusting the switching position of the tap of the discrete action equipment in the preset time period according to the scheduling scheme. The method comprises the steps of extracting data features from load prediction data and active output prediction data through a feature extractor in a discrete action equipment scheduling model, performing data dimension reduction processing and global average pooling processing on the data features to obtain data feature vectors with preset lengths, and determining position prediction values mapped by the data feature vectors through a full-connection layer through an estimation module in the equipment scheduling model, wherein the position prediction values are used for indicating predicted switching positions. The discrete action equipment scheduling model is trained by acquiring historical data of a power distribution network, wherein the historical data comprises historical load prediction data and historical active output prediction data, determining tagged historical data and untagged historical data in the historical data, determining principal component characteristics of the tagged historical data, screening the tagged historical data according to the principal component characteristics to obtain a first training data set, taking the principal component characteristics of the tagged historical data as reference principal component characteristics of the untagged historical data, clustering and screening the untagged historical data according to the reference principal component characteristics to obtain a second training data set, and training the discrete action equipment scheduling model according to the first training data set and the second training data set by adopting an countermeasure generation network. The method comprises the steps of determining a principal component characteristic of tag history data, wherein the principal component characteristic comprises a principal component characteristic vector, the principal component characteristic is used for representing the correlation among all characteristic quantities in the tag data, the principal component characteristic is used for carrying out characteristic value decomposition processing on the principal component characteristic vector, the characteristic value and the characteristic vector corresponding to the characteristi