A DSP Internal Signal-Based Fault Diagnosis Method For Three-Level TNPC Inverter in PMSM Drive System

This paper proposes a fault diagnosis method for three-level T-type neutral-point clamped (3L-TNPC) inverter in permanent magnet synchronous motor (PMSM) drive systems, which utilizes the internal modulation signals of the digital signal processor (DSP), reference voltages, as diagnostic variables. The fault characteristics of the 3L-TNPC inverter and its influence on the modulation signals are analyzed to demonstrate the feasibility of reference voltages as diagnostic variables. Additionally, a fault diagnosis method based on such modulation signals is proposed, which involves generating a two-dimensional coordinate graph and utilizing data-driven methods to diagnose faults with convolutional neural network (CNN) model. Both the online simulation results and offline experimental results verify the effectiveness of proposed method and validate the feasibility of proposed diagnosis strategy.