Offline Robust Model Predictive Control for Lipschitz Non‐Linear Systems Using Polyhedral Invariant Sets

In this paper the concept of maximal admissible set (MAS) for linear systems with polytopic uncertainty is extended to non‐linear systems composed of a linear constant part followed by a non‐linear term. We characterize the maximal admissible set for the non‐linear system with unstructured uncertainty in the form of polyhedral invariant sets. A computationally efficient state‐feedback RMPC law is derived off‐line for Lipschitz non‐linear systems. The state‐feedback control law is calculated by solving a convex optimization problem within the framework of linear matrix inequalities (LMIs), which leads to guaranteeing closed‐loop robust stability. Most of the computational burdens are moved off‐line. A linear optimization problem is performed to characterize the maximal admissible set, and it is shown that an ellipsoidal invariant set is only an approximation of the true stabilizable region. This method not only remarkably extends the size of the admissible set of initial conditions but also greatly reduces the on‐line computational time. The usefulness and effectiveness of the method proposed here is verified via two simulation examples.     

Necessary and sufficient conditions for constraint satisfaction in switched systems using switch‐robust control invariant sets

This paper studies the control of constrained systems whose dynamics and constraints switch between a finite set of modes over time according to an exogenous input signal. We define a new type of control invariant sets for switched constrained systems, called switch–robust control invariant (switch‐RCI) sets, that are robust to unknown mode switching and exploit available information on minimum dwell‐time and admissible mode transitions. These switch‐RCI sets are used to derive novel necessary and sufficient conditions for the existence of a control‐law that guarantees constraint satisfaction in the presence of unknown mode switching with known minimum dwell‐time. The switch‐RCI sets are also used to design a recursively feasible model predictive controller (MPC) that enforces closed‐loop constraint satisfaction for switched constrained systems. We show that our controller is nonconservative in the sense that it enforces constraints on the largest possible domain, ie, constraints can be recursively satisfied if and only if our controller is feasible. The MPC and switch‐RCI sets are demonstrated on a vehicle lane‐changing case study.     

Semi‐global stabilization of discrete‐time systems subject to non‐right invertible constraints

This paper investigates time‐invariant linear systems subject to input and state constraints. We study discrete‐time systems with full or partial constraints on both input and state. It has been shown earlier that the solvability conditions of stabilization problems are closely related to important concepts such as the right invertibility or non‐right invertibility of the constraints, the location of constraint invariant zeros, and the order of constraint infinite zeros. In this paper, for general time‐invariant linear systems with non‐right invertible constraints, necessary and sufficient conditions are developed under which semi‐global stabilization in the admissible set can be achieved by state feedback. Sufficient conditions are also developed for such a stabilization in the case where measurement feedback is used. Such sufficient conditions are almost necessary. Controllers for both state feedback and measurement feedback are constructed as well. Copyright © 2009 John Wiley & Sons, Ltd.     

Stochastic model predictive control for probabilistically constrained Markovian jump linear systems with additive disturbance

Model predictive control (MPC) for Markovian jump linear systems with probabilistic constraints has received much attention in recent years. However, in existing results, the disturbance is usually assumed with infinite support, which is not considered reasonable in real applications. Thus, by considering random additive disturbance with finite support, this paper is devoted to a systematic approach to stochastic MPC for Markovian jump linear systems with probabilistic constraints. The adopted MPC law is parameterized by a mode‐dependent feedback control law superimposed with a perturbation generated by a dynamic controller. Probabilistic constraints can be guaranteed by confining the augmented system state to a maximal admissible set. Then, the MPC algorithm is given in the form of linearly constrained quadratic programming problems by optimizing the infinite sum of derivation of the stage cost from its steady‐state value. The proposed algorithm is proved to be recursively feasible and to guarantee constraints satisfaction, and the closed‐loop long‐run average cost is not more than that of the unconstrained closed‐loop system with static feedback. Finally, when adopting the optimal feedback gains in the predictive control law, the resulting MPC algorithm has been proved to converge in the mean square sense to the optimal control. A numerical example is given to verify the efficiency of the proposed results.     

基于关联矩阵代数运算的逻辑型监控器设计

对于含有不可控变迁的Petri网监控问题,允许状态空间可能需要一组“或”的允许约束来描述,而库所不变量的监控方法[12]只将给定约束转换为单个的允许约束,其监控器将系统行为限制在允许标识状态空间的较小子集内,其限制性过于严格,且该方法无法解决某些监控问题.针对上述问题,给出了一种基于关联矩阵代数运算的约束转换方法,能够...     

基于容许集的变终端约束MPC控制

针对一类输入和状态受约束的离散线性系统,提出一种基于Ⅳ步容许集的变终端约束集模型预测控制方法.首先给出多面体不变集序列作为终端约束集的离线模型预测控制算法,扩大了终端约束集.为进一步扩大初始状态可镇定区域,引入N步容许集,设计了基于容许集的变终端约束集模型预测控制方法.该算法采用离线设计、在线优化方法,实现了系统渐近稳定,不仅降低了在线运算量,而且扩大了初始状态可镇定区域.仿真结果表明了算法的有效性.     

Quadratically saturated regulator for constrained linear systems

The problem of stabilization and null-controllability of open-loop unstable discrete-time multi-input systems with constraints on the inputs and the controls is addressed in this paper. First necessary and sufficient conditions for solvability of the problem are derived. They guarantee the existence of a linear controller leaving the state constraint set for the closed-loop system positively invariant. An optimal control law is computed, and the admissible set of initial conditions is given such that along trajectories of the closed-loop system the state and input constraints are satisfied. Then the domain of feasible initial conditions is enlarged using a saturating control if such is feasible     

一类非线性系统最大可控不变集求解

针对非线性系统线性化在状态约束下最优鲁棒控制求解问题,提出了一种基于混合系统的非线性系统最大鲁棒控制不变集的方法.对于一类非线性系统通过平衡点线性化的方法转化为多模态的混合系统,并进行了混合逻辑动态模型(MLD)的建模,在不变集基本理论的基础上,通过多参数规划的混合整数规划(MIQP)的方法迭代求解最大可控不变集,并求得不变集内的最优控制器,解决系统的状态约束问题.通过一个非线性系统的实例进行建模、仿真,证明了本方法的可行性.     

Exact determinations of the maximal output admissible set for a class of nonlinear systems

This paper is concerned with obtaining necessary and sufficient conditions for fulfilling specified state and control pointwise-in-time constraints against a certain class of nonlinear dynamics. The results are generalizations of the maximal output admissible set theory to the case of nonlinear systems. Main contribution of the present paper is to propose explicit algorithmic procedures to determine the maximal output admissible set. Another contribution is that we discuss on a finite determinability of the maximal output admissible set for nonlinear systems. Some relations between observability of nonlinear systems and finite characterizations of the maximal output admissible set are clarified.     

Discrete-time reference governors and the nonlinear control of systems with state and control constraints

Discrete-time, linear control systems with specified pointwise-in-time constraints, such as those imposed by actuator saturation, are considered. The constraints are enforced by the addition of a nonlinear ‘reference governor’ that attenuates, when necessary, the input commands. Because the constraints are satisfied, the control system remains linear and undesirable response effects such as instability due to saturation are avoided. The nonlinear action of the reference governor is defined in terms of a finitely determined maximal output admissible set and can be implemented on-line for systems of moderately high order. The main result is global in nature: if the input command converges to a statically admissible input and the initial state of the system belongs to the maximal output admissible set, the eventual action of the reference governor is a unit delay.     

An efficient method for stochastic optimal control with joint chance constraints for nonlinear systems

Stochastic model predictive control hinges on the online solution of a stochastic optimal control problem. This paper presents a computationally efficient solution method for stochastic optimal control for nonlinear systems subject to (time‐varying) stochastic disturbances and (time‐invariant) probabilistic model uncertainty in initial conditions and parameters. To this end, new methods are presented for joint propagation of time‐varying and time‐invariant probabilistic uncertainty and the nonconservative approximation of joint chance constraint (JCC) on the system state. The proposed uncertainty propagation method relies on generalized polynomial chaos and conditional probability rules to obtain tractable expressions for the state mean and covariance matrix. A moment‐based surrogate is presented for JCC approximation to circumvent construction of the full probability distribution of the state or the use of integer variables as required when using the sample average approximation. The proposed solution method for stochastic optimal control is illustrated on a nonlinear semibatch reactor case study in the presence of probabilistic model uncertainty and stochastic disturbances. It is shown that the proposed solution method is significantly superior to a standard random sampling method for stochastic optimal control in terms of computational requirements. Furthermore, the moment‐based surrogate for the JCC is shown to be substantially less conservative than the widely used distributionally robust Cantelli‐Chebyshev inequality for chance constraint approximation.     

A robust economic MPC for changing economic criterion

This note presents a robust economic model predictive control controller suitable for changing economic criterion. The proposal ensures feasibility under any change of the economic criterion, thanks to the use of artificial variables and a relaxed terminal constraint, and robustness in presence of additive bounded disturbances. The resulting robust formulation considers a nominal prediction model and restricted constraints (in order to account for the effect of additive disturbances). The controlled system under the proposed controller is shown to be input‐to‐state stable in the sense that it is asymptotically steered to an invariant region around the best admissible steady state. An illustrative example shows the benefits and the properties of the proposed controller.     

Design of an RMPC with a time‐varying terminal constraint set for tracking problem

This paper presents a robust model predictive control algorithm with a time‐varying terminal constraint set for systems with model uncertainty and input constraints. In this algorithm, the nonlinear system is approximated by a linear model where the approximation error is considered as an unstructured uncertainty that can be represented by a Lipschitz nonlinear function. A continuum of terminal constraint sets is constructed off‐line, and robust stability is achieved on‐line by using a variable control horizon. This approach significantly reduces the computational complexity. The proposed robust model predictive controller with a terminal constraint set is used in tracking set‐points for nonlinear systems. The effectiveness of the proposed method is illustrated with a numerical example. Copyright © 2015 John Wiley & Sons, Ltd.     

Probabilistic output admissible set for systems with time-varying uncertainties

This paper considers time-varying uncertain constrained systems, and develops a method for computing a probabilistic output admissible (POA) set. This set consists of the initial states probabilistically assured to satisfy the constraint. The time-invariant counterpart has already been investigated in Hatanaka and Takaba [Computations of probabilistic output admissible set for uncertain constrained systems, Automatica 44 (2) (2008), to appear]. We first define the POA set for time-varying uncertainties with finite dimensional probability space. Then, we show that an algorithm similar to Hatanaka and Takaba [Computations of probabilistic output admissible set for uncertain constrained systems, Automatica 44 (2) (2008), to appear] provides the POA set also in the time-varying case, as long as an upper bound of a what we call future output admissibility (FOA) index is available. We moreover present two methods for computing the upper bound of the FOA index: probabilistic and deterministic methods. A numerical simulation demonstrates the effectiveness of our algorithm.     

Construction of invariant polytopic sets with specified complexity

The problem of construction of controlled invariant polytopic sets with specified complexity, for linear systems subject to linear state and control constraints, is investigated. First, geometric conditions for the enlargement of a polytopic set by adding a new vertex, in order to produce a polytopic set of specified complexity, are established. Next, conditions for such an enlargement of controlled invariant sets to preserve the controlled invariance property are presented. The established theoretical results are used to develop methods for the construction of admissible controlled invariant sets with specified complexity. Two numerical examples show how these results can be used for the computation of monotonic sequences of admissible controlled invariant sets of specified complexity.     

Output feedback receding horizon control of constrained systems

This paper considers output feedback control of linear discrete-time systems with convex state and input constraints which are subject to bounded state disturbances and output measurement errors. We show that the non-convex problem of finding a constraint admissible affine output feedback policy over a finite horizon, to be used in conjunction with a fixed linear state observer, can be converted to an equivalent convex problem. When used in the design of a time-varying robust receding horizon control law, we derive conditions under which the resulting closed-loop system is guaranteed to satisfy the system constraints for all time, given an initial state estimate and bound on the state estimation error. When the state estimation error bound matches the minimal robust positively invariant (mRPI) set for the system error dynamics, we show that this control law is time-invariant, but its calculation generally requires solution of an infinite-dimensional optimization problem. Finally, using an invariant outer approximation to the mRPI error set, we develop a time-invariant control law that can be computed by solving a finite-dimensional tractable optimization problem at each time step that guarantees that the closed-loop system satisfies the constraints for all time.     

Robust variable horizon model predictive control for vehicle maneuvering

This paper provides a new method of robust maneuver control with guaranteed finite‐time arrival and satisfaction of constraints, despite the action of an unknown but bounded disturbance. The new method extends the constraint tightening approach to robust model predictive control of constrained linear systems by combining it with a variable horizon. This relaxes the requirement for the target to be an invariant set, which is assumed by many stabilizing MPC formulations but can be restrictive in vehicle maneuvering applications. The target sets for vehicle maneuvers are typically determined by the mission requirements and are not generally invariant sets. The new controller guarantees finite‐time arrival within an arbitrary target set, i.e. not necessarily invariant, and is therefore applicable when the target is predetermined by other factors. Several simulation examples are presented including spacecraft rendezvous control with sensor visibility constraints and UAV guidance through obstacle fields. Copyright © 2006 John Wiley & Sons, Ltd.     

LMI optimization approach to robust decentralized controller design

A new approach for design of robust decentralized controllers for continuous linear time‐invariant systems is proposed using linear matrix inequalities (LMIs). The proposed method is based on closed‐loop diagonal dominance. Sufficient conditions for closed‐loop stability and closed‐loop block‐diagonal dominance are obtained. Satisfying the obtained conditions is formulated as an optimization problem with a system of LMI constraints. By adding an extra LMI constraint to the system of LMI constraints in the optimization problem, the robust control is addressed as well. Accordingly, the decentralized robust control problem for a multivariable system is reduced to an optimization problem for a system of LMI constraints to be feasible. An example is given to show the effectiveness of the proposed method. Copyright © 2010 John Wiley & Sons, Ltd.     

Maximally permissive supervisor synthesis based on a new constraint transformation method

The method is proposed to design the maximally permissive and efficient supervisor for enforcing linear constraints, in which the weights of places are not negative, on ordinary Petri nets with uncontrollable transitions. First, the weakly admissible linear constraint is introduced. Second, a method is proposed to design the monitor place for enforcing a weakly admissible linear constraint on Petri nets. Third, a theorem proving that a linear constraint can be equivalently transformed at an uncontrollable transition into a disjunction of new constraints is proposed. Fourth, using this theorem, an algorithm is presented to equivalently transform a linear constraint, each place weight of which is not negative, into a disjunction of weakly admissible ones. Lastly, the supervisor, which consists of the plant net and a set of monitor places, is designed for the weakly admissible linear constraints calculated by the above algorithm.     

Distributed stochastic MPC for systems with parameter uncertainty and disturbances

A distributed stochastic model predictive control algorithm is proposed for multiple linear subsystems with both parameter uncertainty and stochastic disturbances, which are coupled via probabilistic constraints. To handle the probabilistic constraints, the system dynamics is first decomposed into a nominal part and an uncertain part. The uncertain part is further divided into 2 parts: the first one is constrained to lie in probabilistic tubes that are calculated offline through the use of the probabilistic information on disturbances, whereas the second one is constrained to lie in polytopic tubes whose volumes are optimized online and whose facets' orientations are determined offline. By permitting a single subsystem to optimize at each time step, the probabilistic constraints are then reduced into a set of linear deterministic constraints, and the online optimization problem is transformed into a convex optimization problem that can be performed efficiently. Furthermore, compared to a centralized control scheme, the distributed stochastic model predictive control algorithm only requires message transmissions when a subsystem is optimized, thereby offering greater flexibility in communication. By designing a tailored invariant terminal set for each subsystem, the proposed algorithm can achieve recursive feasibility, which, in turn, ensures closed‐loop stability of the entire system. A numerical example is given to illustrate the efficacy of the algorithm.