基于PCA-SVM方法的点焊质量评估

Quality evaluation of the resistance spot welding based on PCASVM

通过对电阻点焊过程电极位移和动态电阻信号的实时采集和时域特征分析,利用电阻信号动态特征刻画熔核形成不同阶段,从同步电极位移信号中提取9个特征参量建立输入样本数据集.以焊点接头抗剪强度作为焊点质量的评价指标,将PCA(主成分分析)方法与传统的SVM(支持向量机)回归分析相结合,利用PCA方法对支持向量机的输入样本数据集进行主成分分析,消除了输入特征参量间的自相关性,实现数据降维后作为支持向量机的输入,建立焊点质量映射模型.交叉有效性预测结果表明,基于PCA-SVM的算法增强了SVM的泛化能力,比传统的SVM算法具有更高的预测精度.

The electrode displacement and dynamic resistance signals of resistance spot welding process are collected synchronously.Through the time-domain analysis of electrode displacement signal in the welding process,nine characteristic parameters relating to weld quality are picked up to set up a set of data which characterizes the input samples,on the basis of the different phase of nugget forming marked by simultaneous dynamic resistance signal.The principal component analysis(PCA) to remove the self-correlation of input characteristics and realize dimensionality reduction is integrated with the conventional method of support vector machine(SVM),while the shear strength of welded spot was taken as the evaluation index of welded spot quality.The comparison of predicted results under PCASVM and conventional SVM by means of cross-validation test shows that PCA-SVM algorithm improves the generalization ability and the predicted accuracy of SVM method.