多维力传感器的LS-SVM动态逆解耦建模  

Dynamic Inverse Decoupling Modeling of Multi-axis Force Sensor

作　　者：张景柱[1] 柴焱杰[2] 

机构地区：[1]中国电子科技集团公司第五十四研究所 [2]中国人民解放军96275部队

出　　处：《无线电工程》2014年第8期18-20,44,共4页Radio Engineering

基　　金：国家自然科学基金委委员会;中国工程物理研究院联合基金资助项目(10576039)

摘　　要：维间耦合是制约多维力传感器测量精度的瓶颈。针对维间耦合,提出一种可工程实现的采用LS-SVM非线性估计的α阶逆系统解耦方法,给出了具体实现的过程。在对传感器动态特性未知或者不可直接测量获取的情况下,可以实现多维力的解耦;采用多个二阶线性子系统等效为一个MIMO强耦合系统,利用动态标定多维力传感器的实验来捕捉传感器的动态特性;采取积分器来反映逆系统间的动态变化关系;通过对LS-SVM学习网络的训练实现对逆系统相应的静态非线性函数的逼近,实现力传感器的动态解耦。仿真算例验证了该方法解耦的有效性。The coupling of multi-axis force sensor is a key factor to limit its measuring precision.The αth-order inverse system de-coupling approach is newly proposed in the paper.The implementation process of the LS-SVM inverse decoupling for multi-axis force sensor is presented.One strong coupling MIMO system is decoupled into a number of 2nd-order linear subsystems.The dynamic character of force sensor can be caught by carrying out the dynamic calibration experiments.The dynamic relations among inverse systems can be indicated with integrators.The approximation of static nonlinear function corresponding to inverse system is implemented by training the learning network of LS-SVM.Finally,the decoupling of force sensor is realized.Simulation examples verify the availability of the method.

关 键 词：多维力传感器 解耦 LS-SVM 逆系统 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]