基于SVR的多股簧机床张力控制  

Tension Control Based on Support Vector Regression in Stranded Wire Helical Spring Machines

作　　者：刘青[1] 王时龙[1] 尹梦骄 张其[1] 

机构地区：[1]重庆大学机械传动国家重点实验室,重庆400044

出　　处：《计算机测量与控制》2016年第9期99-102,106,共5页Computer Measurement &Control

基　　金：国家自然基金项目(51375508);国家科技支撑计划项目(2014BAF08B02)

摘　　要：针对多股螺旋弹簧加工机床张力控制系统的非线性特性及存在干扰的问题,提出了基于支持向量回归(Support Vector Regression,简称SVR)的有效抑制噪声的控制算法;即在基于SVR的逆动力学建模中采取非线性松弛因子替代线性松弛因子,抑制噪声在建模过程中的影响,提高建模精度;把系统逆模型作为前馈控制器,构成伪线性系统,PID控制器作为反馈控制,组成有效抑制噪声的复合控制算法;实验结果表明,这种改进基于SVR的复合控制方式在多股螺旋弹簧张力控制系统中具有较好的控制精度和鲁棒性。Giving the nonlinear characteristics and interference in the tension control system in Stranded Wire Helical Spring Machines,a control algorithm based on support vector regression（SVR）was proposed to decrease noise.By replacing linear relaxation factor with nonlinear relaxation factor,and making the inverse model of the system as feed-forward controller and PID（Proportion Integration Differentiation）controller as feed-back controller,a compound control algorithm was constituted to decreasing noise.Experiment has demonstrated that the improved control method has a good control precision as well as good robustness in the tension control system in Stranded Wire Helical Spring Machines.

关 键 词：多股螺旋弹簧张力控制系统 支持向量回归 松弛因子 逆模型控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]