自适应直接逆控制仿真研究  

Simulation Research on Adaptive Direct Inverse Control

作　　者：蒋星军[1,2] 周欣然[3] 唐钊轶[2] 

机构地区：[1]北京工业大学计算机学院,北京100022 [2]湖南广播电视大学计算机系,湖南长沙410004 [3]中南大学信息科学与工程学院,湖南长沙410075

出　　处：《计算机仿真》2013年第12期385-389,394,共6页Computer Simulation

基　　金：国家自然科学基金(10971060);湖南省科学技术厅项目(2011FJ6033)

摘　　要：研究被控对象逆模型控制问题,自适应直接逆控制(ADIC)的关键是即时逆模型的准确性和模型在线辨识算法的快速性。针对逆模型辨识问题,设计了稀疏在线无偏置最小二乘支持向量机(SONB-LSSVM),并提出了基于SONB-LSSVM的ADIC算法。在每个控制周期,进行递推学习新样本,并删除与新样本最相似的样本,然后被共享为控制器并用作计算控制量。仿真表明,SONB-LSSVM能及时学习过程逆动态特性,有较强的泛化能力。表明ADIC具有良好的自适应能力和较高的控制精度。The key problem of adaptive direct inverse control（ADIC） is the accuracy of the instant inverse model and the speedability of online identification algorithm. Considering inverse model identification, a sparse online non - bias least square support vector machine（ SONB - LSSVM） was designed, and an ADIC algorithm was proposed utilizing SONB - LSSVM. During per control interval. The SONB - LSSVM was used to study new sample and remove the sample most similar to the new one. Then, it was shared as controller and control signal was calculated through it. Simulation results show that SONB - LSSVM can study dynamic properties of process in time and has better generalization ability, consequently, the ADIC possesses excellent adaptation and control precision.

关 键 词：逆控制 非线性系统 最小二乘支持向量机 学习算法 稀疏性 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP273[自动化与计算机技术—控制科学与工程]