Abnormal Prediction Of High Precision Machining Based On Unsupervised Learning

In the process of high precision machining,a variety of factors such as the geometric parameters of the cutting tool,the vibration of the cutting tool and the vibration of the tool may have a great influence on the surface quality.The monitoring of the surface quality is time-consuming and time-consuming,which seriously affects the processing efficiency and the product qualification rate.This paper systematically combs the modeling process of the high precision machining surface quality detection at home and abroad and the research status of the clustering algorithm in the fault diagnosis.It combines the microscopic surface of the abnormal quality workpiece observed under the microscope,and designs the processing quality anomaly prediction analysis from the characteristics of the high optical processing and the factors that affect the processing quality.On the basis of long time data acquisition,an automatic data interception technology based on short time energy is introduced,and a number of hot issues such as clustering evaluation,such as clustering evaluation index,based on decision drawing,are built on the basis of long time data acquisition.Heart selection,similarity measure and kernel mapping scheme discussed the establishment of the quality anomaly prediction model and the process of parameter optimization.The generalization test of the model and the comparison with the other clustering algorithms were carried out.Finally,the quality anomaly prediction process based on the kernel clustering algorithm was summarized,and the model site was carried out in a factory high light processing workshop.Verify,predict the quality of the sampleThe study of the surface quality prediction model is helpful to reduce the unnecessary manpower and material resources,adjust the processing conditions in time,thus significantly improve the qualified rate of the workpiece,reduce the production cost,and provide a new way of thinking for the unsupervised learning in the field of machine tool condition detection.