基于支持向量机的校直机工件弯曲量测量误差补偿

Measurement Error Compensation of Straightening Machine Workpiece Curvature Based on Support Vector Machines

针对自动校直机在工件弯曲量测量过程中的测量误差问题,提出一种基于支持向量机(SVM)的校直机工件弯曲量测量误差补偿方法。在试验过程中,利用偏摆检测仪将校直机测得的工件弯曲量进行标定,进而获取建模数据。模型训练过程运用交叉验证法对SVM的惩罚参数、核函数参数进行优化选择,构建SVM优化模型。利用SVM优化模型对不同工件同一测量位置的弯曲量测量误差进行预测补偿,通过对比预测补偿效果,SVM模型的补偿效果明显,具有工程应用价值。

A measurement error compensation method of straightening machine workpiece curvature based on support vector machine（ SVM） was presented to solve the workpiece curvature measurement error of automatic straightening machine. In the process of test,the deflection detector was used to calibrate the workpiece curvature measured by straightening machine and the modeling data were obtained. During the modeling process,the optimized parameters of SVM were chosen by cross-validation method. With these optimized parameters,the established model could be used to predict the workpiece curvature measurement error accurately. The SVM model was respectively used to predict the curvature measurement error of different workpiece and the same position. With the prediction results comparison,the compensation effect of SVM model is obvious. Therefore,it is valuable for engineering application.