Simulation of improved BP algorithm in the fault diagnosis of analog circuit

In view of the complexity of analog circuits, to address the drawbacks of the traditional BP algorithm, such as slow speed of constringency and proneness to lapse into local minima, an improved approach based on affixation momentum & factor adjust and the learning rate is proposed and applied to its fault diagnosis. The results show that the model improves the efficiency of fault diagnosis compared with the traditional method. After the data concerning faults is entered into the trained neural network, the types of the faults can be judged based upon the output, and then the accurate localization of the faulty module can be realized. Thus it can be concluded that the improved BP algorithm is of practical value.