Convolution neural network-based transient stability estimation method of power system

Abstract

The invention discloses a convolution neural network-based transient stability estimation method of a power system. According to the method, time domain response information of amplitude and a phase angle of each bus voltage acquired by a WAMS system in the power system is used as original input characteristic, a transient stability estimation model of the power system is built by a convolution neural network and a Dropout technology and by an image pixel processing method, and the model comprises an input layer, a convolution layer, a pooling layer, a full-connection layer and a Softmax output layer. The input characteristic is not needed to be extracted by manual, and transient stability judgment of the power system is directly performed according to the real-time response information obtained by the WAMS system; and by the method, the network training parameter is greatly reduced, the training difficulty is reduced, the training efficiency is improved, overfitting is effectively prevented, and the method has relatively high estimation accuracy and extremely high generalization capability.

Description

technical field [0001] The invention relates to power system transient stability evaluation and deep learning technology, in particular to a power system transient stability evaluation method based on convolutional neural network. Background technique [0002] The stable operation of the power grid is directly related to the production and life of the people and the normal operation of society. With the continuous expansion of the scale of the power system, the development of AC-DC hybrid transmission methods, the increasing number of new equipment and the application of new energy technologies, the operating state of the system is getting closer and closer to its stability limit, and its security and stability problems are becoming more and more serious. Large-scale power outages occur from time to time, and there is an urgent need for a method that can quickly and accurately realize power grid stability assessment. [0003] The time-domain simulation method is intuitive, ...

AI Technical Summary

Problems solved by technology

However, this method has some disadvantages: (1) It requires complicated manual feature extraction work, and bad features will reduce the evaluation accuracy of the model, and even make the model invalid; (2) It is easy to appear overfitting and lead to the generalization ability of the model insufficient

However, the patent does not specify which specific feature variable vectors are extracted from the sample set, and it does not use specific examples for verification, so the method is unclear, which reduces the credibility of the method to a certain extent and affects engineering practice. application

Images