零件表面加工方法的选择是CAPP系统的重要组成部分,利用概率神经网络以逐步分解的方式来实现零件表面加工方法的选择,并给出了网络的构造、训练和组织训练数据的方法。结果表明,利用人工神经网络可以很好地选择零件表面的加工方法,克服了CAPP专家系统存在的工艺知识获取“瓶颈”问题和“无穷递归”、“推理薄弱”等推理过程中出现的问题。 The selection of the surface machining method is an important part of CAPP system. The selection of surface machining method is realized by using the probabilistic neural network in a stepwise manner. The method of constructing, training and organizing the data is also given. The results show that the artificial neural network can select the processing method of the part surface well, and overcome the bottleneck of the process knowledge acquisition and the problems in the reasoning process such as “infinite recursion” and “weak reasoning” existing in the CAPP expert system.