A data-driven method for IGBT open-circuit fault diagnosis for the modular multilevel converter based on a modified Elman neural network
Yang An, Xiangdong Sun, Biying Ren, Hui Li, Mengnan Zhang
Abstract
Modular multilevel converter is widely used in electrical energy with many advantages, its safety and reliability has become a research hotspot. Since the structure of MMC is composed of multiple cascaded sub-modules, including a large number of IGBTs and capacitors. Therefore, fault diagnosis measures must be taken to quickly eliminate the faults. In order to solve this problem, a data-driven method is proposed based on a modified Elman neural network. By comparing the distance Ek between the predicted and true value of bridge arm current, this method can quickly realize fault detection. The original contribution of this paper is using the modified cuckoo search (MCS) to optimize the parameters of Elman model, so as to achieve the optimal balance between fault diagnosis accuracy and diagnosis speed. The simulation results proved that it can quickly detect the open-circuit fault of IGBT by data-driven, and the detection time is about 20 ms.