Grinding Wheel Wear Monitoring Based on the Time Constant and Support Vector Machine

基於時間常數和支援向量機的砂輪磨損監測

遲玉倫(Yulun Chi) ； 顧佳健(Jiajian Gu) ； 李郝林(Haolin Li)

《中國機械工程學刊》 40卷4期 (2019/08) Pp. 341-353

摘要

In cylindrical plunge grinding process, the wheel wear affects the form accuracy and surface quality of the workpiece. Therefore, monitoring the grinding wheel condition plays a key role in the quality of workpiece being manufactured. In this study, different typical sensors in grinding machines such as the dynamometer, acceleration, motor current and Acoustic Emission (AE) are in use for monitoring. The time constants of force, AE and power signals and the Mean Square Deviation (MSD) of accelerometer signals, are first acquired to study the grinding wheel wear process. Considering that it is not always easy to measure the wheel topography and the grinding surface roughness in real grinding process, the Least Square Support Vector Machine (LS-SVM) is introduced to monitor the grinding wheel wear in time to satisfy the grinding stringent requirements. A series of experiments was performed to verify that the monitoring method of wheel wear is effective and repeatable in grinding process control.

關鍵字

grinding wheel wear ； the time constant ； LS-SVM ； wheel condition monitoring

並列摘要

在外圓切入式磨削過程中，砂輪磨損會影響工件的形狀精度和表面品質，線上監測砂輪磨損狀態對於保證工件品質起著關鍵作用。本研究基於不同的感測器信號，如，力、功率、加速度、電機電流和聲發射（AE）等對數控磨床來進行線上監測。通過計算磨削力、AE和功率信號的時間常數以及加速度信號的均方偏差（MSD）來研究砂輪磨損過程。由於實際磨削過程中測量砂輪形貌和磨削表面粗糙度反應砂輪磨損過程較為複雜且難以實施，本文則採用最小二乘支持向量機（LS-SVM）結合採集信號參數來即時監控砂輪磨損，以滿足磨削加工各項指標要求。最後，通過大量實驗驗證了該砂輪磨損監測方法在磨削程序控制中的有效性和可靠性。