Fault identification method of MMC-HVDC based on GRU neural network

Abstract

In order to transport renewable energy cross regions, building a Modular Multilevel Converter-based High Voltage Direct Current (MMCHVDC) system has become one of the important means. It is necessary to build a high precision and fast response fault identification method to ensure the stability of the system. This research proposes a new fault identification technology based on Gated Recurrent Unit (GRU). This method uses single-ended sensors to obtain the current and voltage before and after the fault. Through the trained neural network based on GRU, it can accurately identify the fault type of the MMC-HVDC system. The simulation experiments indicated that the fault diagnosis method of MMC-HVDC system based on GRU can meet the fault identification speed requirement (3 ms), and the accuracy can reach at 98.94%.