Power System Transient Stability Prediction Algorithm Based on ReliefF and LSTM

Abstract

The stability of the power system has great significance to the national economy, and the sudden situation of the power system will bring great losses. The prediction of the transient stability of the power system can be helpful for coping with the sudden situation of the power system. In order to predict the transient stability of power system, this paper proposes an algorithm based on ReliefF and LSTM network. We first use the ReliefF algorithm to filter features to obtain the most relevant ones, and then the optimal parameters of the LSTM neural network are obtained through iteration, and the trained neural network is used to make a transient prediction of the power network. Compared with the traditional method and SVM algorithm, our algorithm is superior in the aspects of efficiency and accuracy.