基于ＰＤＰ的人工神经网格（ＡＮＮ）研究现在已成为热点，本文针对第一代专家系统自身局限性所造成的知识自动获取瓶颈问题，将神经网络模型引入故障诊断专家系统的研究中来，有效地解决了这一问题。同时，还选取较成熟的ＢＰ算法，运用Ｃ语言开发了一个机械故障诊断专家系统。文中还对神经网络存在的问题进行了分析。 The research of artificial neural network (ANN) based on PDP has now become a hot spot. In this paper, the bottleneck of knowledge acquisition caused by the limitations of the first generation expert system is automatically acquired. The neural network model is introduced into the research of fault diagnosis expert system. To solve this problem. At the same time, we also choose a more mature BP algorithm, using C language to develop a mechanical fault diagnosis expert system. The article also analyzes the problems existing in the neural network.