A Study Of Tool Breakage Monitoring In Milling Based On EEMD And IMF Energy Distribution Using Acoustic Emission Signal

Cutting tool condition monitoring, as an essential part of modernmanufacturing systems, plays an important role on improving product quality,decreasing manufacturing cost and increasing production efficiency. Milling,as one of the most commonly used machining methods, is a kind ofinterrupted processing method, in which tool breakage is one of the mainforms of tool failure. Due to the complexity of the milling process, on-linetool breakage monitoring is still not stable and accurate enough to be appliedto the practical manufacturing and has become one of most popular researchtopics in cutting state monitoring field. Among many monitoring methods,acoustic emission (Acoustic Emission, AE) is generally considered as one ofthe most promising monitoring methods for its high sensitivity, installationconvenience and not affecting the manufacturing process. However, the AEsignal generated from cutting process is a kind of non-stationary signals,which means the frequency characteristics change over time. Traditionalsignal processing techniques, such as analysis method based on FourierTransform, appear to be inadequate in dealing with the non-stationary signals.Otherwise, variety of cutting parameters, entry cut and exit cut to theworkpiece happen frequently during milling process, these factors will affectthe AE signal features in some extent, decreasing the stability of tool breakagemonitoring. Therefore, for tool condition monitoring in milling, dealingnon-stationary signals effectively and improving the monitoring systemstability are key technologies.In this paper, a data acquisition and signal processing system wasestablished to obtain AE signal and extract signal characteristics. A series of experiments were conducted to analyze and deeply summarize thecharacteristics of milling AE signal and tool breakage signal. On this basis, anew recognition algorithm was proposed to monitor tool breakage: First, theoriginal signal was low-pass filtered to filter out low frequency noise；second, Ensemble Empirical Mode Decomposition (EEMD) was used toprocess the filtered signal to extract Intrinsic mode functions (IMF), andHuang-Hilbert spectrum was calculated；The third, the energy of each IMFand the sum of energy of all IMFs were calculated to extract IMF energydistribution curve and changing of the signal energy distribution curve can beused to achieve identification of the tool breakage.In this paper, the proposed recognition algorithm was used to identifytool breakage in milling and the results showed that the proposed method caneffectively extract the energy distribution of the signal state and achieveidentification of tool breakage. Considering the variety of cutting parameters,entry cut and exit cut of the tool into workpiece, a set of experiments wereconducted to verify the stability and robustness of this algorithm. Resultsshowed that the recognition method can eliminate the influence from cuttingparameters and other factors. This paper stated a new solution for toolbreakage monitoring, and meanwhile, has important theoretical and practicalsignificance to promote the application and development of non-stationarysignal processing technology...