基于粗糙模糊神经网络的微孔钻削在线监测

Research on application of rough set-based fuzzy neural network to micro-size drilling on-line monitoring

为解决微孔钻削过程中信号特征与钻头磨损状态关系模型难以建立这一问题,提出一种基于粗糙集理论的模糊神经网络关系模型,首先运用粗糙集理论从数据样本中进行规则集约简,使得网络的模糊规则数目减少,克服了当输入维数高时,模糊神经网络模糊规则过多,结构过于庞大的缺点。然后根据这些规则来设计神经网络的结构模型。应用构造好的网络对主轴电机三相电流信号进行实时数据处理,获取隐含微细钻头磨损状态的信息值,对微孔钻削过程进行在线监测实验,结果表明,适当选择监测阈值,可以有效避免微细钻头的折断。

In order to online monitor micro-size drilling,a rough set-based fuzzy neural network system was developed.By applying rough set theory,the reduced rule sets,i.e.fuzzy neural network rules,were obtained from datum samples and then neural network structure model was designed based on those rules.That reduces the amount of network’s fuzzy rules,gets rid of the disadvantages of overmany fuzzy rules and huge structure for high-dimensional input fuzzy neural network.The so constructed fuzzy neural network was used to real-time process 3 phase current data of spindle motor,to obtain connotative wear states information of micro-size drill,and to perform on-line monitoring for micro-size drilling process.The results were shown that when monitoring liminal value was properly selected,micro-size drill breakage will be effectively prevented.