基于专家知识融入的模糊神经元网络结构及在镗削颤振判别中的应用

在制造系统状态监控中采用神经元网络作为模式识别器，已被证明是一种行之有效的方法。但是，由于制造系统中的加工情况复杂、包含的信息量大，所以神经元网络的学习需要大量样本才能保证它的准确性。本文利用模糊集理论，将专家知识转化为神经元网络可直接处理的模糊ｉｆ－ｔｈｅｎ规则，利用专家知识作为典型样本对模糊神经元网络进行训练，这样节省了大量获取样本的时间，同时又不降低网络的准确性。将之应用于镗削加工中颤振的判别，取得了良好的效果。

Recognition of Chatter in Boring Using Fuzzy Neural Network Integrating Expert Knowledge

Neural networks have been proved to be one effective technique for monitoring of manufacturing processes. However, because of the complex of manufacturing system the accuracy of the training of neural networks require large quantities of information. This paper utilizes fuzzy set theory to translate expert knowledge into fuzzy if then rules which can be processed by a fuzzy neural networks. Consequently, the time spent in getting information used to train neural networks can be saved, at the same time the accuracy of the learning of neural networks can also be ensured. This paper employs this method to recognize the chatter in boring, and the results is satisfied.