Research On Big Data Analysis And Prediction Of CNC Milling Tool Wear State

With the continuous development of Internet of Things,big data,artificial intellige nce and other technologies,intelligent manufacturing has become an inevitable trend in the development of machinery manufacturing.CNC machine tool is an important equip me nt in the intelligent manufacturing system.By monitoring the wear status of the tool during the milling process of the machine tool,it can not only ensure the surface quality and machining accuracy of the processed products,but also ensure the safety of the machine tool equipment,reduce production failures,improve production efficiency,and save manpower and resources.Therefore,the intelligent health monitoring of CNC machine tools has important practical significance.In this paper,the artificial intelligence method is used to intelligently evaluate the milling cutter wear status in CNC machine tools,analyze the milling cutter vibration signal,study the milling cutter wear status recognit io n method,and design offline and online milling cutter wear status monitoring systems.The main work of the paper includes:(1)By analyzing the time-domain waveforms of the X,Y and Z directional vibrat io n signals of the machine tool,ten common time-domain features are extracted,and the correlation between the tool wear state and the time-domain features is studied through visualization techniques.The wavelet packet was used to extract the energy characteris t ics of the high and low frequency bands in different states of the tool,and the energy changes of each frequency band under different wear states were compared.The results show that some time-domain characteristics of the vibration signal and wavelet packet energy characteristics can well reflect the wear state of the tool.(2)Tool wear status recognition is a multi-class problem with unbalanced data.Compared with the normal wear status of the tool,the initial wear and severe wear status information is less.If you do not deal with the data set,it is easy to make the model tend to predict categories with more state information,resulting in insufficient model generalization ability and poor prediction accuracy.Therefore,a balanced data set method based on Borderline-SMO TE is adopted,and the balanced data set is more conducive to ensuring the performance and robustness of the prediction model.(3)A hybrid feature selection method is used to extract the optimal feature set from the original feature set and establish a milling cutter wear state prediction model based on the XGBoost algorithm.Experiments show that this method can guarantee the predict io n performance and model robustness of the model.(4)Designed and developed a prototype system for monitoring the wear status of milling cutters,integrating noise reduction of vibration signals,extraction of time domain and energy features,visualization functions,model training,and online prediction of wear status into the system to achieve Online intelligent identification of tool wear status.