Research on fault diagnosis of three-phase inverter based on BP neural network

In order to quickly and accurately locate the fault location of the inverter circuit, reduce the shutdown time and improve the operation efficiency, an open-circuit fault diagnosis method for the switching device of the inverter circuit based on BP neural network was proposed. Firstly, MATLAB is used to model and simulate the inverter circuit, and the characteristics of the fault signal are directly sampled and extracted from the output voltage waveform. Then, according to these characteristics and the needs of fault diagnosis, a three-layer BP neural network model was constructed, and the number and activation function of neurons were determined. The fault characteristic signal is used as the input of the neural network, and the network is trained by the Levenberg-Marquardt algorithm, and finally the trained neural network model is used to realize the fault diagnosis of the inverter circuit. The experimental results show that the method of direct waveform sampling is simple and effective. Able to accurately locate the MOS tube open circuit fault; The proposed fault diagnosis model has a high diagnostic accuracy.