Research On Tool Wear Status Recognition And Wear Prediction Method Based On Fractal Theory

With the development of the manufacturing industry,tool wear monitoring technology has become a key technology to restrict modern automation machine tools and even promote the development of the country's industrialization,and has been paid more and more attention by researchers both here and abroad.The real-time monitoring of the tool state can improve the machining quality of the parts and the machining efficiency of the machine tools,ensure the operation of the system under the optimal parameters,reduce the occurrence of machine tool accidents,and realize the height integration and unmanned of the machine tools.Therefore,the research of tool wear condition monitoring technology is very urgent and important.In order to quantitatively describe the tool wear stage and predict the wear amount,this paper presents a method based on fractal theory for tool wear identification and wear prediction.First of all,the cutting test is arranged in different cutting conditions by orthogonal test.On the basis of the acoustic emission signal,the improved wavelet threshold function is applied to denoise the signal.Secondly,the feature extraction method of tool wear state based on the multifractal theory is proposed.Finally,the least squares support vector machine is used to realize the wear state recognition and the prediction of the wear quantity..The main contents of the research are as follows:?1?A reasonable tool wear test system is set up to collect the signals in the initial based on the tool wear form and process,normal and sharp wear conditions of the tool.The acoustic emission signal collected by the test is taken as the research object,the frequency range of the acoustic emission signal is determined through the spectrum analysis,and then the improved wavelet threshold method is used to denoise the original acoustic emission signal,and the noise reduction results are verified by the improvement of the signal to noise ratio and the mean square error.?2?In view of the unique advantages of fractal theory in characterizing system nonlinear processes,a tool wear feature extraction method based on multifractal theory is proposed.The fractal characteristics,self similarity,scaling invariance and long range correlation are discussed by using the classical multifractal theory and the multifractal detrending wave analysis.The variation of the multifractal spectrum parameters with the change of the wear quantity under different cutting conditions is compared and analyzed,and the multi fractal parameters,which can sensitively characterize the wear state of the tool are selected,and the characteristics of the multi fractal spectrum parameters?₀,??and?35?h?7?q?8?which are sensitive to the wear state of the tool are selected.Then,multifractal feature parameters are verified by scatter plot to characterize the effectiveness of different tool wear stages.?3?A tool wear identification and prediction method based on multifractal feature and least squares support vector machine is proposed.The recognition accuracy of BP neural network,support vector machine and least squares support vector machine is compared.The least squares support vector machine is superior to the other two algorithms and the recognition accuracy is98.33%.Based on the multifractal feature,the tool wear loss after the tool 20s is predicted by using three regression algorithms,and the prediction results of the wear quantity can be effectively used for tool monitoring.Besides,this method can be extended to predict wear volume under different working conditions and has strong practicability.