基于两级参数算法的对偶控制

针对具有未知参数的随机系统，从随机次优的角度出发，将原不可解的动态规划问题转化为优化一个效用函数，该效用函数考虑了输出调节要求和参数学习要求及两者之间的折衷，充分利用了对偶控制较常规自适应控制的优越性．提出利用两级参数算法来最小化效用函数，从而获得控制信号．仿真结果表明，这种控制器具有良好的对偶性质，能得到较好的学习和控制效果．     

基于两级算法的对偶控制

考虑具有未知参数的随机系统的最优控制问题.采用包含新息方差指数项的损失函数优化系统的性能.新的损失函数由两部分组成:第一部分反映了对输出的调节作用;第二部分反映了辩识系统中未知参数应尽可能多地收集系统信息的需求.提出了一种两级优化算法.该算法首先把不可分问题转化为可分的两目标优化问题,再从两目标优化的非劣解集中挑出原问题的最优解.该控制律易于实施且具有对偶特点.仿真结果表明本文所得的控制律的有效性.     

多变量自适应极点配置对偶控制的研究

针对参数未知的多变量差分方程形式的系统,首先利用Kalman滤波器进行参数辨识,根据确定性等价原理对系统进行极点配置;然后利用极点配置得到的非对偶控制器作为标称输入,其对应的输出作为标称输出,进而根据双指标准则进行对偶控制器的设计;最后给出一个仿真实例,验证该算法的可行性和有效性.     

基于蚁群算法的城域交通控制实时滚动优化

为求解城域交通控制实时滚动优化的混合整数规划模型，将滚动优化各阶段的控制变量映射为一个层状构造图，并定义局部启发信息和信息素更新公式，从而应用蚁群算法搜索各路口的最优信号灯相位序列．对算法复杂度作了理论分析，并通过仿真实验与单路口感应式信号控制技术进行比较，结果表明性能较好．     

基于最大互信息指标的对偶控制研究

针对参数未知但恒定的随机系统, 研究了基于最大互信息指标的对偶控制. 运用了Kalman滤波器估计随机系统未知参数的方法; 研究了最大互信息指标所具有的对偶特性, 即跟踪理想的目标以及探测未知参数的不确定性; 采用了两级优化算法获得次优对偶控制律. 算例验证了此算法的有效性和可行性.     

制造系统的递阶滚动优化调度模型

研究的对象是只有一台不可靠（failure-prone)机器的非完全柔性制造系统，该系统能生产多种产品，但在同一时刻只能生产一种产品，并且当机器由生产一种产品向生产另一种产品切换时，需要考虑setup时间及其成本，待决策变量是setup序列及产品生产率，本文基于非完全柔性制造系统的特点，引入递阶层控的思想，采用新的递阶结构框架和阈值控制策略，对问题进行分解，建立了考虑setup时间及成本的递阶流率控制最优化调度模型，并给出了递阶的滚动优化算法，仿真结果表明，这种调度策略更易于工程实现。     

动态环境中多UCAV协同控制的滚动优化

战场环境的动态性和不确定性以及协同控制的复杂性,使得在线多无人作战飞机(UCAV)协同控制难以满足时间敏感性要求。针对多UCAV协同控制中的任务调度问题,设计了滚动优化方法,从任务层次解决了动态环境中的多UCAV协同控制问题。仿真实验和分析表明该方法简单有效,能够应对战场环境中的多种突发情况,具有很好的时间性能。     

强动态Ad Hoc网的拥塞控制:价格协作和滚动优化

Ad Hoc网络存在着无线多跳连接、节点移动这两个本质的特点.前者引起了与固定网络截然不同的信息流竞争新特点,后者导致了网络状态不断发生变化.首先,在采用链路干扰集描述信息流竞争特点的基础上,将小时间段内网络状态不变的拥塞控制问题表达成非线性优化问题;其次,运用基于对偶分解理论的价格协作法PCA(price cooperation approach)求解该优化问题,构建了一个基于链路干扰集的价格框架.同时,运用队列长度监控、邻居集合近似和HELLO捎带信息这3种技术将PCA转化成在实际Ad Hoc网络环境中可实施的协议.另一方面,运用状态检测和滚动优化方法,有针对性地解决Ad Hoc网络状态不确定时变性带来的问题,相应地设计了自适应优化策略AOS(adaptive optimization strategy).MATLAB仿真结果表明,AOS策略比PCA对网络状态的变化具有更好的自适应性能.NS仿真实验结果表明了PCA和PCA AOS在几乎所有的仿真场景和移动环境下,在重要的性能指标,包括吞吐量、丢包率、公平性等方面,都比TCP,ATCP和ATP有了明显的改进.     

对偶自适应控制

本文针对具有不确定性的随机系统，从随机次成的角度出发，分析了对偶自适应控制的基本原理和优化策略，综述了对偶自适应控制的研究现状和应用成果，并探讨了今后的发展趋向。     

基于领域知识的电梯群控系统的智能滚动优化方法

针对电梯客流的随机性和不可预测性,在群控电梯专家系统的基础上,提出了一种智能滚动优化的方法,获得了更好的控制效果。     

Robust one-step receding horizon control for constrained systems

This paper proposes a robust receding horizon control scheme for discrete-time uncertain linear systems with input and state constraints. The control scheme is based on the minimization of the worst-case one-step finite horizon cost with a finite terminal weighting matrix. It is shown that the proposed receding horizon control robustly asymptotically stabilizes uncertain constrained systems under some matrix inequality conditions on the terminal weighting matrices. This robust receding horizon control scheme has a larger feasible initial-state set and a more general structure than existing robust receding horizon controls for uncertain constrained systems under the same design parameters. The proposed controller is obtained using semidefinite programming. Copyright © 1999 John Wiley & Sons, Ltd.     

Dynamic hedging of basket options under proportional transaction costs using receding horizon control

In this article, we develop a semi-definite programming-based receding horizon control approach to the problem of dynamic hedging of European basket call options under proportional transaction costs. The hedging problem for a European call option is formulated as a finite horizon constrained stochastic control problem. This allows us to develop a receding horizon control approach that repeatedly solves semi-definite programmes on-line in order to dynamically hedge. This approach is competitive with Black–Scholes delta hedging in the one-dimensional case with no transaction costs, but it also applies to multi-dimensional options such as basket options, and can include transaction costs. We illustrate its effectiveness through a numerical example involving an option on a basket of five stocks.     

On the receding horizon hierarchical optimal control of manufacturing systems

This paper concerns the development of a hierarchical framework for the integrated planning and scheduling of a class of manufacturing systems. In this framework, dynamic optimization plays an important role in order to define control strategies that, by taking into account the dynamic nature of these systems, minimize customized cost functionals subject to state and control constraints. The proposed architecture is composed of a set of hierarchical levels where a two-way information flow, assuming the form of a state feedback control, is obtained through a receding horizon control scheme. The averaging effect of the receding horizon control scheme enables this deterministic approach to handle random and unexpected events at all levels of the hierarchy. At a given level, production targets to the subsystems immediately below are defined by solving appropriate optimal control problems. Efficient iterative algorithms based on optimality conditions are used to yield control strategies in the form of production rates for the various subsystems. At the lower level, this control strategy is further refined in such a way that all sequences of operations are fully specified. The minimum cost sensitivity information provided in the optimal control formulation supports a mechanism, based on the notion of a critical machine, which plays an important role in the exploitation of the available flexibility. Finally, an important point to note is that our approach is particularly suited to further integration of the production system into a larger supply chain management framework, which is well supported by recent developments in hybrid systems theory.     

Numerical optimization method for HJI equations derived from robust receding horizon control schemes and controller design

This paper addresses how to numerically solve the Hamilton-Jacobin-Isaac(HJI)equations derived from the robust receding horizon control schemes.The developed numerical method,the finite dierence scheme with sigmoidal transformation,is a stable and convergent algorithm for HJI equations.A boundary value iteration procedure is developed to increase the calculation accuracy with less time consumption.The obtained value function can be applied to the robust receding horizon controller design of some kind of uncertain nonlinear systems.In the controller design,the finite time horizon is extended into the infinite time horizon and the controller can be implemented in real time.It can avoid the on-line repeated optimization and the dependence on the feasibility of the initial state which are encountered in the traditional robust receding horizon control schemes.     

Periodic event-triggering in distributed receding horizon control of nonlinear systems

How to efficiently use limited system resources in distributed receding horizon control (DRHC) is an important issue. This paper studies the DRHC problem for a class of dynamically decoupled nonlinear systems under the framework of event-triggering, to efficiently make use of the computation and communication resources. To that end, a distributed periodic event-triggered strategy is designed and a detailed DRHC algorithm is presented. The conditions for ensuring feasibility of the designed algorithm and stability of the closed-loop system are developed, respectively. We show that the closed-loop system is input-to-state stable if the energy bound of the disturbances, the triggering condition and the cooperation matrices fulfill the proposed conditions.     

Nonlinear stochastic receding horizon control: stability,robustness and Monte Carlo methods for control approximation

This work considers the stability of nonlinear stochastic receding horizon control when the optimal controller is only computed approximately. A number of general classes of controller approximation error are analysed including deterministic and probabilistic errors and even controller sample and hold errors. In each case, it is shown that the controller approximation errors do not accumulate (even over an infinite time frame) and the process converges exponentially fast to a small neighbourhood of the origin. In addition to this analysis, an approximation method for receding horizon optimal control is proposed based on Monte Carlo simulation. This method is derived via the Feynman–Kac formula which gives a stochastic interpretation for the solution of a Hamilton–Jacobi–Bellman equation associated with the true optimal controller. It is shown, and it is a prime motivation for this study, that this particular controller approximation method practically stabilises the underlying nonlinear process.     

基于滚动时域优化的无人飞行器轨迹规划

给出了寻求无人飞行器的最优轨迹的一种方法,其问题描述为使飞行器从初始状态飞行到目标状态,同时避免撞到障碍物。基于混合整数规划的滚动时域优化方法用来求解飞行器的轨迹规划问题。给出的仿真结果显示此方法的有效性以及在复杂环境下的可实时计算性。     

鲁棒后退时域控制中HJI方程的数值解法及控制器设计

本文讨论了由不确定非线性系统鲁棒后退时域控制(robust receding horizon control,RRHC)策略导出的Hamilton-Jacobin-Isaac(HJI)方程的求解,提出了一种新的带反曲变换的有限差分算法计算值函数,所提出算法对HJI方程的求解是一种稳定且收敛的算法.同时提出基于边界值迭代的加速过程,加速优化问题的求解,在花费更少计算时间的前提下,提高计算精度.所求得的值函数可直接应用于一类不确定非线性系统鲁棒后退时域控制器的设计,在控制器设计中,传统鲁棒后退时域控制策略中的有限时域被扩展到无限时域,求得的控制器可实时实现,避免对初始点可解性的依赖以及反复在线优化问题.     

On receding horizon feedback control

Receding horizon feedback control (RHFC) was originally introduced as an easy method for designing stable state-feedback controllers for linear systems. Here those results are generalized to the control of nonlinear autonomous systems, and we develop a performance index which is minimized by the RHFC (inverse optimal control problem). Previous results for linear systems have shown that desirable nonlinear controllers can be developed by making the RHFC horizon distance a function of the state. That functional dependence was implicit and difficult to implement on-line. Here we develop similar controllers for which the horizon distance is an easily computed explicit function of the state.     

Periodic use of time-varying state feedbacks for the receding horizon control of bilinear systems

This paper provides a receding horizon control method for SISO bilinear systems in the presence input constraints. Periodically-invariant sets are derived for a bilinear system with respect to a series of time-varying state feedback gains. The dual-mode control strategy is adopted and the periodically-invariant sets are used as target invariant sets. The state feedback gains used to define the target invariant sets are also used to render degrees of freedom to steer the current state into the target set. The region of attraction for the proposed algorithm is enlarged significantly with an extension of the horizon of periodicity while the on-line. Recommended by Editorial Board member Young Soo Suh under the direction of Editor Jae Weon Choi. This project is conducted through the Practical Application Project of Advanced Microsystems Packaging Program of Seoul Technopark, funded by the Ministry of Knowledge Economy. Young Il Lee was born in Korea in 1963. He received his B.Sc., M.S. and Ph.D. in Control and Instrumentation from Seoul National University. He is currently a Professor of the Dept. of Control and Instrumentation, Seoul National University of Technology. He spent two years at Oxford University as a Visiting Research Fellow. Basil Kouvaritakis was born in Athens, Greece in 1948. He was awarded First-Class Honours in Electrical Engineering from the Manchester Institute of Science and Technology, where he also received his Master and Doctorate. He is currently a Professor in Engineering at the Department of Engineering Science and a Tutorial Fellow at St Edmund Hall, Oxford University. Mark Cannon was born in England in 1971. He received M.Eng. and D.Phil. degrees from Oxford University and S.M. from Massachusetts Institute of Technology. He is currently a University Lecturer in Engineering and Fellow of St. Peter’s College, Oxford University.