On-line Monitoring Of Tool Wear For Mass Production

Tool wear has influence on the workpiece machining accuracy, surface quality, and leads to machine vibration, noise and other unpredictable consequences. For many years, there have been various research on the identification of tool wear, but all are in the study stage. The tool wear identification theory, which is applied in practical production, still does't have related report. Existing methods of tool wear monitoring are hardly suitable for mass production of cutting parameter fluctuation. This paper presents a tool wear identify method which based on the analysis of current signals. The main research contents are as follows:According to the characteristics in mass production and the tool wear rule, this paper presents a tool wear identify method, which monitoring drive current signal of spindle motor and through the tool wear rule learning. By determining the experimental program, we finish the experiment design of surface milling tool wear identification in machining center. Then to establish the current signal monitoring experimental system, and complete related experiment.By analysising the current signal using wavelet packet analysis, many eigenvalues of tool cutting current signal are got. Through the analysis of sensitivity between tool wear and eigenvalue, the eigenwalues of high sensitivity are selected to reflect tool wear.Through tool wear rule learning, the tool life model is got using BP neural networks. The tool life quantitative identification is completed by neural network pattern recognition technology. On the late stage of tool life, some of the recognition results error is within 10%, which can both ensure replace the worn tool opportunely and improve tool utilization.By establishing the learning relation model between tool wear value and signal eigenvalue, tool wear value can be identified real-time. Then the workpiece size error cased by tool wear can be compensated. The error of workpiece size after compensation reduced to only 25% compared to that without compensation.