Sound Signal And Vibration Signal Processing For Tool Breakage Monitoring Based On EMD And ICA

Metal-cutting process monitoring is the key technology and theimportant aspect of automation machinery manufacturing. With theimprovement of automation technology FMS and CIMS，the tool conditionmonitoring system need to be more practical and reliable. However, themeasured signals from sensors always contain lots of information relatedto cutting process and unrelated to cutting process. How to obtain thetarget information is the key issue to be solved in the whole cutting processmonitoring. This paper studies the intelligence fusion ways which isapplied to both cutting sound signals and vibration signals in the millingprocess, so as to separate related target condition information in millingprocess. The study possesses theoretical and practical significance ofinformation extraction technique in cutting process monitoring.In this thesis, a multiple-channels signal blind source separationalgorithm is proposed based on EMD and ICA using multi-sensor information fusion technology, the tool breakage monitoring relatedtechnical problems in milling process are analyzed in detail based onsound sensor and vibration sensor which are used for signal acquisition.Firstly, use graphical integrated development environment LabVIEW todevelop the system as data acquisition platforms, in accordance with theexperimental program, collect different multi-sensor signals in the differenttool condition cutting process. Then, extract the eigenvalue related with thetool breakage by comparing the different tool condition signals, Finally,the information related to milling cutter is separated from cutting soundand vibration signals, based on the proposed multiple-channels blindsource separation technique.The study of this thesis demonstrates that the method based On EMDand ICA is capable of simultaneously separating components concerningdifferent cutting condition from mixed cutting sound and vibrationmonitoring signals sources, even if the sources share common frequencyband. This study not only brings new solution for tool conditionmonitoring, but also extends the application range of ICA and EMD.