Short-circuit current null point forecast method based on Bi-LSTM neural network

The difficulty in phase-controlled short-circuit fault current breaking is to forecast the null point of short-circuit current in a quick and accurate manner. To this end, a short-circuit current null point forecast method was researched and proposed based on the bidirectional long- and short-term memory (Bi-LSTM) neural network algorithm. First of all, a Bi-LSTMbased short-circuit current null point forecast model was constructed whose accuracy and effectiveness were assessed by root-mean-square error (RMSE), correlation coefficient R² and other indicators. Then, the accuracy and rapidity of the forecast of short-circuit current null point through Bi-LSTM model were verified by emulation dataset. According to the research findings, the Bi-LSTM-based network presents a good forecast capacity regarding short-circuit current null point, with its sampling time reduced to less than 3 ms and its null point error with ± 0.2ms, and therefore can provide a theoretical support for the phase-controlled short-circuit fault current breaking.