基于最小二乘支持向量机的一类非自治系统自适应控制

Adaptive Control of a Class of Non-Autonomous Systems Based on Least Square Support Vector Machine

提出了一种基于最小二乘支持向量机(LS-SVM)的一类不确定非自治系统自适应控制器设计方法.该方法基于最小二乘支持向量机来估计对象的部分未知非线性项,并给出了最小二乘支持向量机权向量和偏移值的在线学习规则.利用李亚普诺夫理论严格证明了整个闭环系统的跟踪误差、控制器参数以及最小二乘支持向量机权参数和偏移值的一致最终有界.此控制方法可以保证对象在线稳定地跟踪任何光滑的目标轨迹,仿真结果表明了此控制方法的可行性和有效性.

A design method of adaptive controller is proposed based on least square support vector machine （LS-SVM） for a class of uncertain non-autonomous systems, in which LS-SVM is adopted to estimate the partially unknown nonlinear items in the object, and an on-line learning rule for the weight vectors of LS-SVM and the bias is derived. Lyapunov theory is used to strictly prove the uniform ultimate boundedness of the tracking errors of the whole closed-loop system, the controller parameters, the weight parameters and the bias of LS-SVM. The proposed control scheme can track any smooth target trajectory steadily and online, and the simulation result shows its effectiveness and feasibility.