Modal Analysis Of Variable Machine Tool Position Based On Clustering Method

The performance of the machine tool are closely related to the dynamics of the machine tool,and the dynamics of the machine tool are closely related to the component position of the machine tool.In the process of machining,the change of the component position can't be avoided,so it is very important to study the modal parameters under the change of component position.The traditional research on machine tool dynamics under the change of machine state is to manually analyze the response data of different states one by one,identify the modal parameters in each component position,and compare the modal parameters obtained by manually analysis to find changes of the modal parameter.It is extremely complicated to identify the modal parameters from the response signal.The identification process is greatly interfered with by harmonics.The process of determining the modal parameters is subjective and the analysis results of different personnel may have human error.The results of manual analysis are difficult to become the unified standard for machine tool.The analysis process is often repeated,and a large number of repeated work is needed in the study of machine tool modal parameter under state changes,so the manual analysis is inefficient.In this paper,a new modal analysis method applied to variable machine tool position based on clustering is proposed.Main work includes(a)the characteristics related to the modal parameters are extracted from the machine response signal,and the clustering results are used to reflect the changes of machine tool modal parameters.Combined with the theory of traditional modal research,this paper proposes the frequency response function as the modal characteristic of the machine tool.The simulation results show that the clustering result of the frequency response function can reflect the modal parameter change.(b)In the analysis of the measured data,the power spectrum is used to replace the frequency response function,and the clustering results are also good.However,which modal parameter changes can't be obtained from the clustering results only based on power spectrum.(c)This paper proposes a preprocessing method to normalize the frequency response function,which stresses the natural frequency information and shields the amplitude of the signal.The result of the clustering after signal normalized is only related to changes in the natural frequency.In this paper,the multi-component multi-point response data clustering experiment is designed according to the dynamic characteristics of the machine tool to verify the effectiveness of the normalization method.(d)The clustering method is used to analyze the modal parameter changes under different position conditions of the machine tool,and the manual analysis The results were compared to verify the effectiveness of the clustering method.Compared with the traditional manual analysis method,the signal-processing process in clustering method is not affected by the machine harmonic,so there is no manual intervention and no artificial error.In addition,the new method is high efficiency,and the clustering results directly show the modal parameter changes in different position,which can be used as a standard method to characterize the modal parameter in variable component position.