基于支持向量机方法建立超声振动磨削放电加工预测模型
Establishing Ultrasonic Vibration Grinding EDM-assisted Processing Prediction Model Based on Support Vector Machines

针对采用机器学习理论建立超声振动磨削放电加工模型时存在试验样本数量少、预测量数值变化波动大的问题，提出利用支持向量机方法建立加工指标预测模型的方法。以超声振动磨削放电加工SiCp/Al为例，利用正交试验获取学习样本数据，采用MATLAB软件建立超声振动磨削放电加工SiCp/Al工艺指标的支持向量机预测模型，并利用该模型预测零件表面粗糙度和加工速度两项工艺指标。结果表明：支持向量机模型得到的工艺指标预测值与试验值具有较好的一致性，最大相对误差不超过12%，预测值精度较高，所建立的超声振动磨削放电加工工艺指标的支持向量机预测模型是可靠且有效的。
Aim at the numerically fluctuated prediction of the ultrasonic vibration grinding assisted by EDM(electric discharge machining) in application of the models is built based on the machine learning theory, a processing prediction model based on the support vector machine(SVM). Taking the ultrasonic vibration grinding of SiCp/Al assisted by EDM as an example, learning samples with orthogonal tests and establish the prediction model is constructed based on SVM on the MATLAB platform. Then the model is used to predict two indicators such as surface roughness and processing velocity. The research results show that the prediction results of the processing prediction model are in accord with the test results. When the maximum relative error is less than 12% and the prediction value being highly accurate, the prediction model is reliable and effective