Research On Complex Fault Diagnosis Of CNC Machine Tool Feed System Based On Multi Source Information Fusion

CNC machine tool is the most typical mechanical and electrical integration equipment.It is the most commonly used in modern industrial production.The feed system as an important part of CNC machine tools,its performance parameters and feed efficiency directly affect the machining accuracy of CNC machine tools and production quality.The rolling bearing and ball screw pair are the two main parts of the feed system,which play the role of support and transmission.If the two faults,it will affect the transmission efficiency of the entire transmission system,and then have a huge impact on the CNC machine,and even damage.Therefore,to carry out on NC machine tool feed system performance testing and fault diagnosis research,for ensuring the safety of CNC machine tools,reliable,efficient operation has a very important significance,also has the very high practical value to the actual engineering application.In this paper,the research status of complex fault at home and abroad is analyzed based on the rolling bearing and ball screw pair in the feed system of CNC machine tool.In view of the limitation of single sensor information and single information source,a complex fault diagnosis method based on multi source information fusion is proposed.Collect the multi-source information of the complex fault by using the vibration sensor,noise sensor and temperature sensor and programming with MATLAB and LabVIEW platform to realize pattern recognition and fault diagnosis for complex fault diagnosis.Analysis of rolling bearing and ball screw fault mechanism and failure forms,ten kinds of complex fault types studied are established by using NI data acquisition card and five external sensors,and the design of MATLAB LabVIEW programming based on the fault signal acquisition and storage.Using the EEMD method in time domain,frequency domain and time domain to extract fault signal.The SLLE dimension reduction method is introduced to simplify and filter the large feature vectors,which can lay the foundation for improving the speed and accuracy of the diagnosis.On the basis of analyzing the characteristics of the vibration signal,the method for the separation of complex fault vibration signals based on EEMD-FastICA is proposed.Through the MATLAB platform,the simulation model of the rolling bearing compound fault is established,and the results are compared with the test results.The characteristics of fault data structure are analyzed,the complex fault diagnosis method based on manifold clustering analysis is presented,and the relationship between data is described by manifold distance.And compared with other methods,it is found that this method is more intuitive and simple,and the diagnostic accuracy is higher.The intelligent diagnosis system based on BP,RBF and GRNN three kinds of neural networks is established,and the characteristic value of the composite fault filter is characterized by feature level fusion.The recognition performance of the three networks is studied through experiments.The results show that the GRNN neural network has advantages in the complex fault diagnosis,and the diagnostic results are more accurate.Based on the characteristics of the uncertainty of the composite fault information,the improved D-S evidence theory is put forward.The concept of clustering coefficient is introduced to improve the theory of evidence,and it is applied to the composite fault diagnosis of rolling bearing and ball screw pair,and the decision level fusion is realized.Experiments show that this method can reduce the conflict rate between the evidence bodies,and can effectively improve the accuracy of the complex fault diagnosis.