关键词: 逆变电路 小波分析 神经网络 DSP 故障诊断

Abstract:

For the characteristrics of nonlinearness of inverters which cannot use an accurate mathematical model for fault diagnosis, a novel algorithm based on wavelet analysis and neural networks is presented to detect and diagnose inverters. A nonlinear MATLAB simulation model of a three-phase sine-PWM inver ter is set up. The three-phase voltage acts as the fault information, drawing out the fault characteristics by wavelet analysis as the inputs of neural   networks to locate the inverter bridge with the switch fault. Finally,a simple judgement strategy for locating the fault switches based on the symmetry  of the three-phase fault voltage is proposed. An on-line fault diagnosis system based on DSP is designed. The test resuits prove that the system feature s better effect on the fault diagnosis,and the method is practical and feasible.

Key words: inverter, wavelet analysis, neural network, DSP, fault diagnosis