A Model Prediction Method for Open Circuit Fault Diagnosis of Multi-tube Power Devices based on Common Mode Voltage

As we all know, the power component fault of motor drive system will affect the reliability and stability of the system. Traditional fault detection methods mostly use Space Vector Pulse Width Modulation(SVPWM) to explore single power device fault or two power device faults of three-level inverter, but SVPWM cannot detect three power device faults. In this paper, a model prediction method based on common mode voltage(CMV) for open circuit fault diagnosis of multi-tube power devices is proposed. Firstly, the motor drive system is a hybrid system composed of discrete switching signal and continuous current state. The mathematical model of three-level inverter is established and different fault types are defined. Secondly, an improved midpoint voltage vector model predictive control strategy is proposed to avoid the traditional modulation strategy. A dual-vector cost function based on midpoint voltage vectors is constructed, and the optimal vectors of the remaining sectors are determined and the driving signals are output after considering different fault types. Then the CMV is selected as the new characteristic signal, and the fault diagnosis of the three power devices is realized precisely by judging the amplitude of the CMV corresponding to the optimal vectors. Finally, the effectiveness of the proposed method is verified by simulation and experiment results.