Research On Open-circuit Fault Diagnosis And Fault-tolerant Control Of MMC Sub-module

Modular multilevel converter(MMC)is widely used in flexible DC transmission,new energy grid connection and other fields because of its advantages of good output waveform quality,good scalability,high efficiency and good switching characteristics.With the increase of the number of MMC bridge arm cascade sub modules,and affected by mechanical stress and chemical stress,the sub modules are prone to failure.If they are not detected and removed in time,it will cause great harm to the system.This paper takes modular multilevel converter as the research object,focusing on the open circuit fault of sub module,analyzes and studies the fault characteristic analysis,fault diagnosis and location and fault tolerant control.This paper analyzes the topology,mathematical model and sub module working principle of MMC,introduces the basic control strategy,modulation strategy and circulating current suppression strategy of MMC,and analyzes the sub module open circuit fault characteristics,which provides a theoretical basis for the research of sub module fault diagnosis and location and fault tolerance control.A new fault diagnosis method of fuzzy sliding mode observer based on bridge arm current is proposed.In this algorithm,the switching function of traditional sliding mode observer is replaced by exponential function,and the sliding mode gain is dynamically adjusted by fuzzy control algorithm,which can weaken the chattering problem caused by high frequency switching signal and improve the accuracy of fault diagnosis of bridge arm.According to whether the current of MMC bridge arm is intermittent or not,the open circuit fault is divided into two types.The energy entropy of wavelet transform is extracted by wavelet transform of capacitor voltage of sub module,and the position of open circuit component is determined according to the change rate of wavelet energy entropy.BP neural network is used to transform the sub module fault location problem into a classification and identification problem.By coding the bridge arm sub module,the wavelet packet node energy of the sub module capacitor voltage decomposed by wavelet packet is input into the neural network for training.Genetic algorithm is used to optimize BP neural network to make the fault location more accurate.The fault-tolerant control strategy of sub module based on bridge arm energy prediction is adopted to ensure the stable operation of the system after the sub module fault is removed.This measure calculates the reference value of the capacitor voltage of each bridge arm sub module in real time by predicting the energy stored in the bridge arm sub module,so as to determine the switching of the sub module,so as to suppress the circulating current and harmonics in the MMC system after the sub module fault is removed.Based on the MATLAB / Simulink simulation platform,the proposed fault diagnosis,fault location and fault tolerant control method are verified.The results show that the proposed method is feasible and effective.