基于2D理论的点对点综合预测迭代学习控制  

Point-to-point integrated predictive iterative learning control based on two-dimensional theory

作　　者：洪英东 熊智华[1] 江永亨[1] 叶昊[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《控制与决策》2018年第3期431-438,共8页Control and Decision

基　　金：国家自然科学基金项目(61473162)

摘　　要：基于二维系统综合预测迭代学习控制(2D-IPILC)方法,结合轨迹更新策略研究点对点跟踪问题的控制算法.该算法既能够充分利用点对点问题在非跟踪点的自由度,也可以通过引入模型预测控制来提高时间轴的抗干扰能力.由于轨迹更新中引入时变参数,该2D模型为时变2D模型,因此分析状态转移矩阵特性和系统全响应,进而采用2D理论分析算法的收敛性和收敛条件,并分析参数对控制效果的影响.相比固定轨迹算法,该算法的收敛速度更快,稳定性比直接型优化算法更好.最后通过仿真实例验证了所提出算法的效果.A two-dimensional based integrated predictive iterative learning control（2 D-IPILC） method is applied to the point-to-point tracking problem with the trajectory updating strategy. The 2 D-IPILC method can fully take advantages of the freedom brought by the untracked points, and can improve the capability of dealing with the disturbance in the time domain by combining the feature of model predictive control. As time-varying parameters are involoved in the updating trajectory scheme, the model of the 2 D-IPILC method is also time-varying. Therefore, the characteristics of time-varying transition matrix and full reponse are analyzed in detail. Furthermore, the convergence principle and conditions of the2 D-IPILC method are proved based on the 2 D theory, and the effects of the control parameter setting are also discussed.The 2 D-IPILC method can get faster convergence than the fixed trajectory methods, and is more robust than the direct optimization algorithm. Finally, these advantages are demonstrated by numerical simulation.

关 键 词：迭代学习控制 2D系统理论 点对点跟踪 轨迹更新 预测控制 时变模型 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]