Research On Chatter Detection Method For Micro-milling Based On SVM

Chatter is a common type in the processing of micro milling.Once appear in the process of machining will leads to reduced machining precision of the workpiece,increase the surface roughness,and even reduce the service life of cutting tool and machine tool processing,such not only can cause a lot of economic losses,there are serious casualties.Many scholars carried out extensive research for method of chatter detection in micro milling.In order to meet the needs of society and under the premise of guarantee the precision of the workpiece as far as possible to reduce economic loss.We need to clear the characteristics of chatter signal and detect the chatter signal accurately for suppress chatter.In this paper,the main work is research on the chatter signal recognition and detection,the paper's content includes:(1)The background,significance and current situation of chatter detection in micro milling are introduced.The development of sensors in chatter detection and chatter detection in micro-milling is described.Researches and improvements of the chatter detection technology of scholars at home and abroad are summarized in recent years.(2)A suitable sensor is chosen to collect the machining signal in micro-milling,and the characteristics of signals collected by different sensors are also compared.(3)In order to obtain the chatter signal,an unidirectional flexible fixture is designed,and the frequency response function of fixture is obtained and the experimental parameters are choosed.The collected acceleration signal is analyzed and discussed,and the characteristics of the chatter signal are obtained.(4)The improved Hilbert transform method(HHT)was used to extract the signal characteristics in time-frequency domain.The characteristics of each characteristic parameter in steady machining and unstable machining are analyzed and compared.In the time-frequency domain analysis,time-frequency entropy is selected as the signal feature.(5)The algorithm of Support Vector Machine(SVM)based on time-frequency entropy is proposed to recognize the state of micro-milling,and the optimal parameters c and g are set up.The prediction of the machining state of micro-milling is realized.(6)The state of the machined surface during chatter is analyzed and the roughness parameters of the surface were analyzed and compared.The results showed that the surface roughness of the workpiece increased greatly when the flutter occurred.