A Data-Driven Fault Diagnosis Method with Dimension Reduction Capability for Inverter Open-circuit Fault of Multiphase Drive Systems

Abstract

In recent years, multiphase motors have been widely used due to their excellent performance, with fault-tolerant operation being one of the main reasons. Accurate and fast fault diagnosis is the prerequisite for fault-tolerant operation. In this paper, a data-driven scheme based on Random Vector Functional Link (RVFL) network with data dimension reduction capability is proposed to diagnose the power switch open-circuit faults of the multiphase inverter. The enhanced fast Fourier transform (FFT) considering the rotational speed of the motor is adopted to extract the phase current features precisely, and then the dimensions of data are reduced through the ReliefF algorithm. At last, the faults are detected and diagnosed by the RVFL network. Simulation experiments and comparison analysis are conducted on a six-phase permanent magnet synchronous motor (PMSM) drive system, which demonstrate that the proposed method can effectively extract fault features and achieve high accuracy while reducing the computational effort.