最小二乘支持向量机在多层金属膜厚度检测中的应用  

Application of Least Square Support Vector Machine in Measuring the Thickness of Multiple Layers Metal Film

作　　者：李玉军[1] 梁琨[2] 郭会军[1] 

机构地区：[1]西安理工大学自动化与信息工程学院,陕西西安710048 [2]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《西安理工大学学报》2010年第3期345-350,共6页Journal of Xi'an University of Technology

基　　金：陕西省自然科学基金资助项目(SJ08F31)

摘　　要：利用多频激励的单一电涡流传感器对工件表面两层金属膜厚度及传感器探头与被测工件之间间隙进行检测,采用多项式拟合法及最小二乘支持向量机技术对系统分别建模。实验结果表明,采用多项式拟合建立的预测模型可以使得传感器的最大位移灵敏度系数由1.03mm-1下降至0.27mm-1,对系统性能有一定的改善;采用最小二乘支持向量机技术建立的预测模型可以使传感器的最大位移灵敏度系数趋于零。传感器测量间隙对铝膜厚度及铜膜厚度测量的交叉敏感得到了有效的抑制,从而提高了系统模型的预测精度。因此采用最小二乘支持向量机建立的回归模型具有较好的预测能力及分析优势。A single eddy-current sensor is used to measure the thickness of multiple layers of metal film and the clearance between the sensor and the work piece. Both polynomial fitting and Least Square Support Vector Machine （LS-SVM） are used to establish the model of system respectively. The experiment results show that the forecasting model established using polynomial fitting can reduce the max displacement sensitivity coefficient of the sensor from 1.03mm^-1 to 0.27mm^-1. The system performance can be improved to a certain extent. The results also show that the forecasting model establihed using LS-SVM can reduce it to zero. So that the cross-sensitivity between the measuring clearances with the thickness of aluminum foil and copper foil is controlled efficiently with LS-SVM, whereby the forecasting precision of the system model is improved. Therefore, the regression model established by LS-SVM has better forecasting ability and analysis advantage.

关 键 词：电涡流传感器 最小二乘支持向量机 数据拟合 交叉敏感 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TP181[自动化与计算机技术—控制科学与工程]