Milling Cutter Wear Fault Monitoringbased On Multiple Machine Learning Algorithms

In the process of transition to digitalization,networking and intelligence,the traditional machinery processing industry has gradually exposed many problems in production,management and other aspects,which has promoted the increasing automation technology of mechanical processing.The intelligent diagnosis of CNC machine tool faults can effectively ensure the quality of the workpiece and the safety of CNC machine tools.The tool is a cutting tool in the process of CNC machine tool processing.The tool wear in the high-speed machining state for a long time is more serious,and the sharp wear or damage of the tool will cause the machine tool to be damaged or the workpiece is unqualified.Therefore,it is important to accurately judge the tool wear.Through the fault diagnosis of the cutting tools of CNC machine tools,the problems arising from the transformation of CNC machine tools to intelligence can be effectively solved.First collect the milling cutter acceleration signal and cutting force signal of the CNC machine tool,and perform the time domain,frequency domain and time frequency domain analysis on these signals to obtain the corresponding eigenvalues,and then use the kernel principal component analysis(KPCA)to characterize the obtained eigenvalues Screening process to get the feature matrix after dimension reduction.On the basis of the above,the extracted new feature vector is used as the input vector of the machine learning algorithm,and the tool fault is intelligently diagnosed through the machine learning model.In this paper,for the fault diagnosis of CNC machine tools,two methods are used to identify the wear state of the milling cutter of the CNC machine tool and predict the wear of the milling cutter as follows.(1)Recognition of the wear state of the milling cutter of CNC machine tools.For the recognition of the milling cutter wear state,because the SVM has good small sample and nonlinear data processing capabilities,and the SVM parameters have a greater impact on the experimental results,so the particle swarm optimization algorithm is used to optimize the SVM parameters,and then the milling cutter wear state assessment At the same time,the BP_AdaBoost algorithm is also used to evaluate the wear state of the milling cutter.The research shows that the BP_AdaBoos algorithm is more accurate in evaluating the wear state of the milling cutter of the CNC machine tool.(2)Prediction of the amount of wear of milling cutters of CNC machine tools.For the prediction of milling cutter wear,SVR can overcome the model's unconstrained data distribution and the influence of abnormal points on prediction,and SVR is greatly affected by parameters.For this reason,first use particle swarm optimization to optimize SVR parameters,And then predict the amount of milling cutter wear.At the same time,the long-term and short-term neural network(LSTM)algorithm is also used to predict the tool wear.The research shows that the LSTM algorithm is more accurate in predicting the wear of CNC machine tools.In this paper,a variety of machine learning algorithms are used: optimized SVM model,BP_AdaBoost,optimized SVR model,and long-term and short-term memory neural network(LSTM)to effectively identify the wear state of the cutter and effectively predict the wear amount of the cutter Establish the basis for predicting CNC machine tool faults.