Optimization of grinding Process by Means of In-process Measurement of grinding wheel Surface

本研究的目的是通过对砂轮表面的过程测量来优化磨削操作。项目大纲如下。(1)车轮现场监测系统建设。(2)测量环法对应的车轮表面表征。(3)车轮表面的过程评价。砂轮表面表征参数分为静态参数和动态参数两大类。在后处理静态评价中，采用针形法和显微法直接测量车轮表面。用触笔轮廓仪和数码显微镜分别测量了车轮表面的周向轮廓线和二维图像。在这里，根据任意定义的标准来识别磨料颗粒和切削刃。该方法可测量磨粒的晶粒密度、尖端密度、连续尖端间距、尖端比和微形貌，精度较高。该技术可以相对容易地测量砂轮形貌，从而可以通过判断磨削条件的适宜性和重新布置砂轮的时机来控制磨削操作。另一方面，在过程动态测量中，通过分析磨削区域内砂轮与工件相互作用产生的磨削声音来评价砂轮状态。通过适当修整形成砂轮表面的若干参考条件，并利用神经网络学习算法对参考砂轮表面发出的磨削声的频谱进行判别。整体识别率至少为60%。该网络还将车轮磨损识别为车轮表面磨尖变平，这与较低修整进料时产生的车轮磨损相当。因此，该系统可以即时识别车轮表面的差异。少

The purpose of this study is to optimize the grinding operation by means of in-process measurement of the wheel surface. The outlines of the project are as follows.(1) Construction of the in-situ wheel monitoring system.(2) Characterization of the wheel surface corresponding to the measure ring method.(3) In-process evaluation of the wheel surface.Parameters for characterizing grinding wheel surface are classified into two categories : static and dynamic parameters. In the post-process static evaluation, the wheel surface is directly measured by means of the stylus profilometry and the microscopy. The circumferential profile trace and 2D-image of a wheel surface is measured by the stylus profilometer and the digital microscope, respectively. Here abrasive grains and cutting edges are identified according to arbitrarily defined criteria. The grain density, cutting-edge density, successive cutting-edge spacing, cutting-edge ratio and micro-morphology of abrasive grains can be measured wi … More th a good degree of accuracy. This technique enables to measure the wheel topography relatively easily so that it has the possibility to control the grinding operation by judging the suitability of the grinding condition and the timing of re-arrangement of the wheel.In the in-process dynamic measurement, on the other hand, the wheel condition is evaluated by analyzing the grinding sound, which is generated by the wheel-work interaction in the grinding zone. Several reference conditions of the wheel surface are formed by the appropriate dressing and the frequency spectrums of grinding sound emitted from the reference wheel surface are discriminated by the neural network learning algorithm. The overall recognition rate is at least 60%. The network also recognize the wheel wear as the wheel surface with flattened grain tips, which is equivalent to that of the wheel generated with lower dressing feed. Accordingly this system can recognize instantaneously the difference of the wheel surface. Less