Multi-fault Integrated Classification and Identification Method Based on LSTM

There are many devices in the distribution system, transformers, circuit breakers and various types of power equipment are more susceptible to the external environment, resulting in complex and diverse fault types, which seriously affects the safety and reliability of the power grid. The existing protection equipment lacks a unified algorithm framework to identify and protect multiple faults in the distribution network. Based on this, this paper proposes a comprehensive fault identification model based on long and short term memory (LSTM) neural network algorithm. Firstly, this paper collects experimental data for typical faults such as short circuit, surge, residual current, overload, over-temperature, over-voltage and under-voltage. Secondly, the typical features of various types of faults are analyzed and calculated accordingly. On this basis, a comprehensive fault classification and recognition model combining LSTM and threshold judgment is proposed. Finally, the validity of the model is verified by the data set collected and constructed before. The comprehensive fault identification model proposed in this paper can accurately identify the short circuit, surge, residual current and other faults in the distribution network lines, and then provide targeted protection for the distribution network lines by controlling the circuit breaker to open and close, ensure the normal and stable operation of various power equipment, and effectively improve the safety and reliability of the distribution network.