Research On Condition Monitoring Technology Of CNC Machine Tool

Manufacturing is the foundation of the national economy,and CNC machine tools are the core of basic manufacturing capabilities.Failure of its key components can directly reduce the processing efficiency of the machine tools,so the condition monitoring and performance evaluation of its key components are particularly important.In the production process,once a fault is found,it will be immediately shut down for maintenance,resulting in abundant data in the healthy state and a lack of data in the faulty state.The lack of fault data restricts the development of tool condition monitoring and performance evaluation technology.Therefore,studying how to identify tool health and evaluate tool performance under the condition of lack of fault data is of great significance to improve the level of manufacturing.Condition monitoring and evaluation of CNC machine tool are focus of the article.Details are as follows.1.The mechanism of tool wear is described,and the method of dividing the APMT 1135 milling cutter wear stage and the standard of dullness are given according to the national standard.An experimental platform is designed and built,42 tools wear experiments are carried out,and multi-sensor data during the tool life cycle is collected,laying a data foundation for the development of condition monitoring and performance evaluation technology research.2.Aiming at the problem of low recognition accuracy of the minority samples when the intelligent fault diagnosis method processes imbalanced mechanical fault data,a classification method named Deep Cost Adaptive Convolutional Network(DCACN)is proposed.Different misclassification costs are set for different categories and continuously updated adaptively,which improves the recognition rate of minority samples while ensuring the recognition accuracy of the majority samples.3.Aiming at the problem of the lack of tool failure state data,a performance evaluation method is proposed,which only needs the data of the normal state of the tool.Reconstruction error of RBM is used to construct health indicator,which uses the degree of data deviation from normal state to characterize tool state.Then combined the health indicator and simplified tool model,and tool's remaining life is predicted by particle filter algorithm,which realizes the evaluation of the tool performance status under the condition of lack of failure data.4.The tool condition monitoring and evaluation system is developed.The hardware system is built,and the software is modularized based on the algorithm described above.Furthermore,the front-end and back-end of the monitoring system are developed based on the B/S architecture,realizing real-time condition monitoring and performance evaluation of the tool.