State monitoring and fault diagnosis of the PWM converter using the magnetic field near the inductor components

Abstract

This paper proposes a new fault diagnostic method for the PWM converter. A loop magnetic near field probe is used to detect the magnetic field near the inductor components in the PWM converters, and the measured waveform is utilized as the diagnostic criterion. The features of the waveform are extracted by the Fast Fourier Transform, and the interested low and high order harmonic components are used to classify the states of the converters. The low order harmonic components are classified by the Back Propagation Neural Network and the high order harmonic components are classified by the simple mathematical method. Finally, by compromising both the two results, the detailed diagnostic conclusions are obtained.