小波神经网络在模拟电路故障诊断中的应用

模拟电路故障诊断有故障字典法、故障参数识别法、故障验证法。小波神经网络可避免MLP等神经网络结构设计的盲目性，具有逼近能力强、网络学习收敛速度快、参数的选取有理论指导、能有效避免局部最小值问题等优点。随着小波理论和神经网络理论的不断发展，小波神经网络将日益成熟应用于模拟电路故障诊断领域。

Application of Wavelet Neural Networks in Analog Circuit Fault Diagnosis

The analog circuit fault diagnosis includes fault dictionary method, fault parameter identification method, and faults validate method. The wavelet neural network is totally without the blindness of MLP constructer design. It includes performances such as strong approaching, high speed of network learning convergence, selecting parameters based on theory guidance, and avoiding local minimum. Based on developing of wavelet theory and neural network theory, the wavelet neural network is applied to the analog circuit fault diagnosis.