基于支持向量机的激光焊接过程的非线性辨识  

NONLINEAR IDENTIFICATION OF ARC WELDING PROCESS BASED ON SUPPORT VECTOR MACHINE

作　　者：毛剑[1] 刘玉生[1] 

机构地区：[1]四川大学电气信息学院,四川成都610065

出　　处：《计算机应用与软件》2009年第1期16-17,43,共3页Computer Applications and Software

基　　金：国家自然科学基金国际合作项目(60540420641)

摘　　要：针对激光焊接过程非线性系统建模困难的问题,研究基于支持向量机的非线性系统回归建模方法。支持向量机由核函数与训练集完全刻画,进一步提高支持向量机性能的关键是针对给定的系统设计恰当的核函数。用改进的核函数,对具有典型非线性特性的焊接过程进行辨识。仿真结果验证了该方法的有效性。To solve the problem that the nonlinear system of laser welding process is difficult in modelling, in this paper it studied the modelling of nonlinear system regression based on support vector machine. SVM is completely characterized by kernel function and training set, the key to further enhance the performance of support vector machine is to choose appropriate kernel function for the given system. An improved kernel function was applied in the identification of a welding process with typical nonlinear character. The simulation results show that the method is effective.

关 键 词：支持向量机 核函数 非线性系统 焊接过程 辨识 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP13[自动化与计算机技术—控制科学与工程]