The best way to conference proceedings by Francis Academic Press

Web of Proceedings - Francis Academic Press
Web of Proceedings - Francis Academic Press

Design and Research of Electronic Circuit Fault Diagnosis Based on Artificial Intelligence

Download as PDF

DOI: 10.25236/wccece.2018.07


Wang Yongdong

Corresponding Author

Wang Yongdong


Neural network fault diagnosis is based on the direct waveform analysis of on-line fault diagnosis of power electronic circuits. Taking the three-phase rectifier circuit as an example, the waveform output when the circuit breaks down is analyzed. The samples made from the sampling data of the fault waveform are used to train the neural network, and the trained neural network is used to diagnose the fault. Simulation results show that this method is effective. The actual operation of the power electronic circuit shows that most of the faults are manifested by the damage of the power switching device, of which the open circuit and the direct power switching device are the most common. There is a big difference among power electronic circuit fault diagnosis and general analog circuits, digital circuit fault diagnosis, and fault information exists only in the failure to power within a few dozen milliseconds. Therefore, it is necessary to conduct real-time monitoring and online diagnosis. This paper mainly studies the application of neural network theory for power electronic circuit fault diagnosis using neural network learning ability, so that the fault waveform and the relationship between the cause of the fault through the neural network to learn in the structure of its preservation, and then will learn Neural network for fault diagnosis. Neural network can know the cause of the failure through the analysis of current voltage waveform, in order to achieve fault online automatic diagnosis. The following is an example of fault diagnosis of inductive load three-phase rectifier circuit, and the fault diagnosis method of neural network based on waveform direct analysis is studied.


Fault diagnosis, neural network power, electronic circuit.