Cutting Process Monitoring Based On Blind Source Separation

Cutting process monitoring is one of the key problems for modern machining systems to automatization, digitization and intellectualization. Among most monitoring methods, the spindle motor power monitoring is considered to be one of the most promising method for shop floor applications because it is relatively simple and its mounting hardly affects the machining operations. However, the measured power signals from sensors always contain lots of information related to cutting process and unrelated to cutting process. How to obtain the target information is the key issue to be solved in power monitoring and the whole cutting process monitoring. In this thesis, blind source separation is applied to power signals in order to separate related target condition information in milling power signals. Therefore, the study possesses theoretical and practical significance of information extraction technique in cutting process monitoring.In this thesis, a single-channel signal blind source separation algorithm is proposed based on wavelet transform and independent component analysis. First, selection of numbers of wavelet scale levels is analyzed based on pseudo frequency and the selection strategy of wavelet scale levels of power signals in milling is obtained. Wavelet coefficient in each scale level can be considered to be a blend of source signals. Then, independent component analysis is used to separate these wavelet coefficients signals. Simulation shows that this algorithm can effectively separate individual source signals from a single mixed signal. Finally, the related information in power signals, such as source signal related to spindle condition and source signal related to milling cutter, is separated based on the proposed single-channel blind source separation technique.The study of this thesis demonstrates that blind source separation can separate power signals in milling. It brings a new solution to cutting process monitoring and possesses theoretical and practical significance in spread application and development of blind source separation.