基于支持向量机的高速电火花小孔加工穿透检测技术

Breakthrough Detection Technology of High Speed EDM Drilling Based on Support Vector Machine

为解决高速电火花小孔加工穿透检测技术的难题,分析并提出采用支持向量机分类算法对穿透检测特征量进行处理,将小孔穿透检测看作数据分类问题,利用支持向量机构建穿透检测的SVC模型。通过实验分析和数据处理,选择脉冲宽度、脉冲间隔、加工电流、有效放电频率和电极进给速度作为SVC模型的五个输入量,输出为是否发生穿透。穿透检测的SVC模型采用径向基核函数,通过对样本的训练学习,实现了对小孔穿透瞬间的可靠辨识。实验结果表明:在正常加工情况下,支持向量机辨识模型对不同的工件厚度均能在电极穿透工件瞬间进行有效辨别。

In order to solve the breakthrough detection problem in high speed EDM drilling,the support vector machine classification algorithm is used to deal with the feature of breakthrough detection. The breakthrough detection is regarded as the data classification problem,and the support vector classification（SVC） model of breakthrough detection is achieved by using the support vector machine. The input and output of the SVC model are chosen by the experimental data analysis. The output variable is whether the electrode penetrates the workpiece,and the input variables are pulse width,pulse interval,machining current,effective discharge frequency and electrode feed rate. The SVC model of breakthrough detection uses radial basis function as kernel function,and through the training of sample data,the reliable identification of penetration is realized. The experimental results show that under normal processing condition,the SVC model of breakthrough detection has high accuracy and reliability.