At present, the intelligent fault diagnosis technology is widely used in the diagnosis of large equipment such as emergency diesel engine and steam turbine in power equipment, power supply system. However, the application of this technology in the more sophisticated circuit board is very rare. The diagnosis object chosen in this paper is circuit component, with the help of PSPICE software, the simulation modeling, fault setting and fault modeling of the card are carried out, and the waveform Library of the output pin of the card is obtained. Using waveform recognition diagnosis method to select fault waveform, the fault waveform data are input into the built BP neural network diagnostic model, and the fault feature knowledge base is obtained. It provides an effective solution to the fault location and maintenance of the circuit board in practical use.