基于模糊神经网络的数控机床刀具磨损量预测研究  

Research on Tool Wear Prediction of NC Machine Tools Based on Fuzzy Neural Network

作　　者：徐磊 XU Lei(Department of Mechanical Engineering,Nanjing Vocational Institute of Mechatronic Technology,Nanjing 211306,China)

机构地区：[1]南京机电职业技术学院机械工程系,江苏南京211306

出　　处：《机械制造与自动化》2024年第5期87-91,共5页Machine Building & Automation

摘　　要：为实现机床高效运转,提出基于模糊神经网络的数控机床刀具磨损量预测方法。根据小波包分析方法划分频带为多个层次,将信号特征作为选取频谱与频段的参考依据,提取刀具磨损特征。融合模糊逻辑与神经网络,模糊处理数据样本,得到带有模糊规则的数据形式,通过隶属函数描述网络输入项与输出项间的映射关系,模糊处理输入变量,利用模糊神经网络的5个网络层完成数控机床刀具磨损量预测。经比对方均根误差及仿真实验结果证明:所建模型能够有效适应磨损量的变化情况,准确预测出任何工况阶段的刀具磨损量,具有较强的可靠性与准确性。The prediction of tool wear of NC machine tools based on fuzzy neural network is proposed for the efficient operation of machine tools.By the wavelet packet analysis method,the frequency band is divided into multiple levels,and the signal characteristics are used as the reference basis for selecting the frequency spectrum and frequency band to extract the tool wear characteristics.Fuzzy logic and neural network are fused and data samples are fuzzily processed,and the data form with fuzzy rules is obtained.The mapping relationship between network input and output items is described by membership functions.Aariables are input in fuzzy processing,and the five network layers of fuzzy neural network are applied to complete the prediction of tool wear of digital control machine tools.The comparison of the simulation results such as root mean square error proves that the established model can effectively adapt to the change of wear amount and accurately predict the tool wear amount It has strong reliabiliky and accuracy.

关 键 词：数控机床 刀具磨损 磨损量预测 模糊神经网络 模糊神经元 

分 类 号：TP206[自动化与计算机技术—检测技术与自动化装置]