基于拟Broyden法的非线性系统参数优化迭代学习控制  

Parameter-optimal Broyden-like iterative learning control for nonlinear systems

作　　者：逄勃[1] 邵诚[1] 

机构地区：[1]大连理工大学先进控制技术研究所,辽宁大连116024

出　　处：《大连理工大学学报》2013年第4期586-592,共7页Journal of Dalian University of Technology

基　　金：国家自然科学基金资助项目(61074020)

摘　　要：为解决动力非线性系统跟踪控制问题,将拟Broyden法和参数优化迭代学习控制方法结合,即利用拟Broyden算法对系统雅可比矩阵进行迭代近似计算,通过参数优化对学习因子进行优化,提出了一种新的具有单调收敛特性的迭代学习控制算法.该算法不仅能够简化传统牛顿法中对系统雅可比矩阵求逆计算所带来的复杂性,而且从理论上证明了其具有单调递减的特性和全局收敛性.仿真结果表明,该算法能够精确地跟踪给定输入目标,具有实施简便和单调超线速收敛的特点.To solve the tracking problems of dynamic nonlinear system, combining Broyden-like method with a parameter-optimal iterative learning control scheme, a new monotonically convergent iterative learning control algorithm is proposed. The Broyden-like method is used to iteratively calculate the approximation of the Jacobian matrix, and the parameter-optimal method is used to optimize the learning gain of the algorithm. Compared with traditional Newton-method, the proposed algorithm can avoid the calculation of the inverse of the Jacobian matrix. The property of global and monotonic convergence of the algorithm is also proved theoretically. A simulation is carried out to verify the theory, and the numerical results demonstrate that the algorithm can accurately track the given target, and has the properties of simple implementation and monotonic superlinear-velocity convergence.

关 键 词：迭代学习控制 Broyden法 单调收敛 参数优化 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]