基于SMO-SVM的单点金刚笔钝化监测

Tool Wear Monitoring of Diamond Single-point Dresser Based on SMO-SVM

针对单点金刚笔在砂轮修整过程中易于钝化且难以检测的问题,使用支持向量机建立智能模型。为了得到建立模型所需的样本库,使用小波包分析等方法在线提取修整时声发射信号中的特征信息,并引入钝化平台直径定义钝化临界值。模型本身选用基于串行优化算法的支持向量分类机,使用交叉验证法搭配遗传算法以达到优化模型参数的目的。实验结果表明,该模型在分类精度和计算时间上均优于一般的智能模型,可以有效地监测金刚笔的钝化。

An intelligent monitoring model was proposed based on support vector machine to solve the problem of identifying the wear of diamond single-point dresser in the dressing process of grinding wheel. To obtain the required samples for modeling, wavelet packet analysis was used to extract the feature informations from acoustic emission signals during the dressing process, and the diameter of wear platform was employed to define the threshold of dresser wear. Besides, for improving the prac- ticability of the model, a SOM method was applied to train the support vector classifier, the parame- ters of the model were selected by using genetic algorithm as well as cross validation method. Experimental results show that the model has higher performance than general intelligent model, and can monitor the wear of the dresser effectively.