Research On Tool Condition Monitoring Method Of Milling Process Based On Stationary Subspace Analysis

As one of the most widely used equipment,NC milling machine directly affect the implementation of intelligent manufacturing.As the end part of the milling machine,the cutter is one of the key factors for the success of milling.Thus,it is very important to timely and effectively identify and monitor the milling tool which is the most vulnerable part of the milling machine.This paper studied the tool condition monitoring(TCM)technology of milling process,and provides some effective information for tool condition evaluation and maintenance decision.This paper firstly reviewed several technologies of TCM in milling from the aspects of sensing technology,feature extraction and monitoring model,summarized the classical signal processing methods,and analyzed the theoretical principle and algorithm process of stationary subspace analysis method to lay a solid theoretical foundation for the subject.Aiming at the characteristics of small sample size and time-varying signal for milling TCM,this paper proposed a new multi-sensor tool condition monitoring method in milling based on stationary subspace analysis.The stationary subspace analysis method was applied to transform the multi-dimensional signals to stationary source and non-stationary source,then ten dimensionless time-frequency indexes of the non-stationary signal were extracted to train least squares support vector regression to obtain a tool wear estimation model.The analysis and comparison of experimental data verified the feasibility and effectiveness of the proposed method.For the TCM under single sensing,this paper proposed a new single-sensor milling tool condition monitoring method with stationary subspace analysis and phase space reconstruction technology.Firstly,the phase space reconstruction technique was used to extend the one-dimensional signal into multi-dimensional signal.Subsequently,stationary subspace analysis was used to extract stationary and non-stationary sources from the extended multi-dimensional signals.10 dimensionless parameters in the time-frequency domain for non-stationary source were calculated to train the multi-classes least squares support vector machine.Finally,through the experimental platform of the four-blade end milling cutter in NC milling machine,the monitoring experiment of tool wear condition is carried out.The analyses of the experimental results show that the proposed method is more effective than other several methods.