An Intermittent IGBT Trigger Fault Diagnosis method with Multi-Objective Pulse Prediction

Insulated Gate Bipolar Transistor (IGBT) is a crucial component widely used in power conversion equipment, whose reliability investigation is of great significance to guarantee the stability of system operation, improve the equipment lifespan and decrease the failure rates. This paper concentrates on the abnormality between the IGBT module and the controller, investigating the diagnosis method for its control pulse intermittent fault. The proposed method takes the single-phase pulse rectifier as the research object, combining the NARX structure with adaptive exogenous inputs and the RVFL neural network to establish a data model with multi-objective prediction, which achieves the simultaneous monitoring of the four IGBT pulses in the rectifier, and the prediction accuracies can all exceed 98%. In addition, the fault severity of impulsive stochastic is clearly demonstrated through the development of periodic failure assessment criteria. the method is evidenced to be effective in detecting the occurrence of intermittent triggering control faults of IGBT modules after an experiment, and achieving the security maintenance of the system for early faults.