Tool wear is inevitable in actual manufacturing, especially in extreme processing conditions for milling difficult-to-cut materials. With the degradation of cutting performance, the undesirable tool status could trigger instability in machining process that causes negative effect on machined surface and dimension accuracy of the precision components. Therefore, tool condition monitoring for machining process is imperative. With distinctive characteristics on easy acquisition and non-interference in machining process, the power consumption of machine tool has been proved to be a reliable indicator for reflecting the changes of cutting process in industry. Thus, the present study proposes a novel method for monitoring tool condition using net cutting power consumption. The result indicates that the developed approach performs better than traditional approach which uses original total power to predict tool wear. This can be attributed to the changes of net cutting power consumption is more relevant with tool wear evolution. Furthermore, predicting tool wear under variable cutting parameters is achieved using Gaussian process regression, because it has good generalization performance and also explains the uncertainty of prediction results in the form of probability. This study reveals that low-cost sensor like power meter, can be used as an important supplement to monitor tool condition in the industry, and also provides a research basis for predicting tool wear under different cutting conditions.

No abstract