本文提出了用多小波神经网络诊断模拟电路故障的新方法。根据多小波的多分辨率分析思想,构造了一种多小波神经网络,其激励函数为具有紧支撑集、对称性和正交性的多尺度函数和多小波函数。用多小波神经网络和单小波神经网络对实例电路进行故障诊断,仿真结果表明,与单小波神经网络相比,多小波神经网络不仅学习收敛速度快,而且能够更准确地对模拟电路进行故障诊断。 In this paper, a new method to diagnose the fault of analog circuit by multiwavelet neural network is proposed. According to multi-wavelet multi-resolution analysis, a multi-wavelet neural network is constructed, whose excitation functions are multi-scale functions and multiwavelet functions with tight support set, symmetry and orthogonality. The multi-wavelet neural network and single wavelet neural network are used to diagnose the fault of the circuit. The simulation results show that compared with single wavelet neural network, multi-wavelet neural network not only has fast learning convergence, but also can accurately diagnose analog circuits .