Recurrent graph convolutional network system for power grid transient stability evaluation

Abstract

The invention discloses a recurrence plot convolutional network system for power grid transient stability evaluation. The system comprises a space extractor, a time sequence extractor, a global time pooling and a classifier of a tax protection library. A new graph relation matrix of a convolution layer is designed by taking line admittance as reference, a special graph convolution network is formed, and power grid node state information transmission and fusion based on a topological relation are realized by adopting a graph convolution algorithm; a space extractor containing a special graph convolution network is cascaded with a time sequence extractor containing a long short term memory (LSTM) network, and a time pooling layer is designed behind the time sequence extractor to aggregate prediction results of a plurality of time steps so as to form a recursive depth graph convolution model suitable for power grid stability evaluation. Spatial layer and time layer feature aggregation andfeature extraction of power grid state quantity data can be sequentially realized, so that a power grid transient stability evaluation system established based on the method can adapt to the change of a power grid topological structure, excellent generalization ability is shown, and an evaluation result is accurate.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a recursive graph convolution network system used for power grid transient stability evaluation. Background technique [0002] The artificial intelligence method based on machine learning establishes a power system transient stability evaluation model, which can realize fast online stability judgment. It takes sample learning as the core, and mines the mapping relationship between power grid state quantities and stable results through offline learning, without the need to build a large power grid mathematical model. At present, most of the power grid stability assessments based on artificial intelligence use shallow machine learning models, such as support vector machine (support vector machine, SVM), multilayer perceptron (multilayer perceptron, MLP), etc., which need to rely on expert experience and manual feature extraction. It is easy to ignore important information, a...

AI Technical Summary

Problems solved by technology

However, these general models cannot effectively consider the influence of power grid topology, nor can they take into account the integration and integration requirements of time correlation and spatial correlation between huge power grid state information.

This makes it difficult for the power grid transient stability evaluation system based on the conventional deep learning model to maintain a high evaluation accuracy after the power grid topology changes, and the topological changes of the actual power grid due to maintenance or failure occur frequently

Images