Chaos Prediction of Power Systems by Using Deep Learning

Abstract

Ensuring the stability of power systems is an important issue that should be considered in order to ensure the social and economic development of a country. Therefore, predicting the chaotic behavior of power systems in order to develop protection measures and keep power systems stable is vital. In this paper, a deep learning algorithm was proposed to predict the chaotic behavior of power systems by using deep long short-term memory (DLSTM) networks, which have two forms: deep long short-term memory with static scenario (DLSTM-s) and deep long-term memory with dynamic scenario (DLSTM-d). The genetic algorithm was used to optimize the hyperparameters of the networks. Then, taking interconnected power systems as an example, the effectiveness of the proposed DLSTM network was verified via numerical simulation. Finally, the experimental results of the DLSTM network were compared with those of the echo state network, multi-recurrent neural network, deep gated recurrent unit, and long short-term memory. Experimental results illustrated that a trained DLSTM network can predict the chaotic behavior of power systems by using the time series data of a single state variable. Moreover, the DLSTM-s network proposed in this paper can achieve competitive prediction performance compared with other baseline methods.