Milling Wear Monitoring Study Based on Cutting Power

Multi-parameter signals of the milling tools wear are pretreated. Using feature fusion pattern, structural levels project of the signal level, the model level, the characteristic level and class structure are established. The optical control strategies of the tools compensation system is proposed through training samples fuzzy wavelet neural network approach system, so that tool wear is estimated. This contrast results from the experiments, the forecast accuracy of the artificial neural network models is in the context. When the tool wear is small reversely, and the relative error of the model is larger. Because the effective power caused by the tool wear is small, and the measured results are influenced easily by the other factors. When the tool wear is large, the relative error of the model is small. Results of experiment show that the model is applicable to the monitor milling tool wear in different milling process and can more accurately monitor acute wear.