Analog Circuit Fault Diagnosis Based on RBF Neural Network Optimized by PSO Algorithm

The present paper proposes a fault diagnosis methodology of analog circuits base on radial basis function (RBF) artificial neural network trained by particle swarm optimization (PSO) algorithm. Using the appropriate stimulus signal, fault features are extracted from efficient points in frequency response of the circuit directly, and then a fault dictionary is created by collecting signatures of different fault conditions. Trained by the examples contained in the fault dictionary, the RBF neural network optimized by PSO has been demonstrated to provide robust diagnosis to the difficult problem of soft faults in analog circuits. The experimental result shows that the proposed technique is succeeded in diagnosing and locating faults effectively.