介绍了双向联想记忆（ＢＡＭ）学习算法神经网络和误差后向传播（ＢＰ）神经网络在数控系统故障诊断中的应用实例。这两种神经网络在数控系统故障诊断方面都有着重要的应用价值，都能实现故障的联想、容错、自学习等功能。但因为各自具有的特点不同，又使得ＢＰ网络更适于故障样本多的情况，而ＢＡＭ则适于要求对错误的故障征兆进行纠错的故障诊断系统 The application examples of BAM learning algorithm neural network and error back propagation (BP) neural network in CNC system fault diagnosis are introduced. Both of these neural networks have important application value in the fault diagnosis of numerical control system and can realize the functions of fault association, fault tolerance and self-learning. But because of their different characteristics, but also makes the BP network is more suitable for the failure of a large number of samples, and BAM is suitable for fault indications for error correction fault diagnosis system