Study On State Recognition Of CNC Machining Process Based On Multi-sensor Information Fusion

As the core of machine tool intelligence research,CNC machine tool condition monitoring plays an important role in ensuring the safe and stable operation of machine tools,improving processing quality and production efficiency.However,the CNC machining process state recognition as a key part of the machine condition monitoring,has been the focus of research at home and abroad.In this paper,the processing state of CNC machine tool is taken as the research object,and advanced signal processing and pattern recognition methods are used to study the state recognition of CNC machine tools,which is of great significance to the exploration of machine tool processing rules,cutting parameter optimization,processing state control methods and strategies.In this paper,starting from the research on the system structure and working principle of CNC machine tool,and clarifies the types of monitoring signals and sensor specifications for state recognition research of machine tools by analyzing the characteristics of different monitoring signals of machine tools.Then,through the study of the vibration state of the key parts of the machine tool,a vibration state recognition method of the machine tool spindle based on low dimensional features is proposed.On the one hand,the local discriminant wavelet packet transform is used to extract the state characteristics of the spindle vibration signal,which solves the problems of determining the decomposition layer,selecting the basis wavelet function and decomposing the redundant signal in the wavelet packet transform.On the other hand,the principal component analysis is performed on the extracted feature vector to reduce the feature dimension and calculation cost,and the supervised learning method is used to realize the state recognition based on the low-dimensional feature information.In order to overcome the uncertainties and limitations of the single-sensor information analysis in the research of machine tool key parts status recognition,the architecture and methodology of multi-sensor information fusion are studied at the data level and feature level respectively,and a machine tool processing condition recognition method based on multi-sensor information fusion is proposed to solve the influence of insufficient state information on recognition results.In order to verify the feasibility of the proposed different multi-sensor information fusion methods,specific blade processing experiments were designed,and then the effectiveness of the multi-sensor information fusion method was verified through the multichannel vibration and sound data of the processing process.Finally,a multi-sensor information fusion machine tool processing state recognition system was developed based on the mixed programming method of LabVIEW and MATLAB.While retaining the respective advantages of the two software,and solves the problems of complex process and heavy workload caused by the independent operation of data acquisition and data processing system.The multi-sensor information fusion method proposed in this paper for the state recognition of CNC machine tools comprehensively utilizes the state information of different sensors.Compared with single-sensor information analysis,it improves the accuracy of state recognition and can be effectively applied to the state monitoring of CNC machine tools.