LMI-based robust model predictive control and its application to an industrial CSTR problem

In this paper, robust model predictive control (MPC) is studied for a class of uncertain linear systems with structured time-varying uncertainties. This general class of uncertain systems is useful for nonlinear plant modeling in many circumstances. The controller design is characterizing as an optimization problem of the “worst-case” objective function over infinite moving horizon, subject to input and output constraints. A sufficient state-feedback synthesis condition is provided in the form of linear matrix inequality (LMI) optimizations, and will be solved on-line. The stability of such a control scheme is determined by the feasibility of the optimization problem. To demonstrate its usefulness, this robust MPC technique is applied to an industrial continuous stirred tank reactor (CSTR) problem with explicit input and output constraints. Its relative merits to conventional MPC approaches are also discussed.     

Incorporating state estimation into model predictive control and its application to network traffic control

Model predictive control (MPC) is of interest because it is one of the few control design methods which preserves standard design variables and yet handles constraints. MPC is normally posed as a full-state feedback control and is implemented in a certainty-equivalence fashion with best estimates of the states being used in place of the exact state. This paper focuses on exploring the inclusion of state estimates and their interaction with constraints. It does this by applying constrained MPC to a system with stochastic disturbances. The stochastic nature of the problem requires re-posing the constraints in a probabilistic form. Using a gaussian assumption, the original problem is approximated by a standard deterministically-constrained MPC problem for the conditional mean process of the state. The state estimates’ conditional covariances appear in tightening the constraints. ‘Closed-loop covariance’ is introduced to reduce the infeasibility and the conservativeness caused by using long-horizon, open-loop prediction covariances. The resulting control law is applied to a telecommunications network traffic control problem as an example.     

Stochastic MPC with inequality stability constraints

Stochastic uncertainty is a common feature of many control engineering problems, and is also present in a wider class of applications, e.g. finance and sustainable development. Recent work proposed a constrained MPC approach that took explicit account of the distributions of uncertain model parameters but used terminal equality constraints to ensure stability. The present paper reformulates the problem in order to relax the stability constraints by invoking appropriate terminal inequalities. The application of the proposed strategy and its advantages over earlier work are illustrated by means of a numerical example.     

Improved robust model predictive control with structured uncertainty

In this paper, a dilation of the LMI characterization is presented to address constrained robust model predictive control (MPC) for a class of uncertain linear systems with structured time-varying uncertainties. The uncertainty is described in linear fractional transformation (LFT) form. By introducing slack variables and using parameter dependent Lyapunov functions, the design conservativeness is reduced compared with other existing MPC approaches. The proposed approach is applied to an industrial CSTR benchmark system to demonstrate the merits of our proposed solution.     

Robust output-feedback model predictive control for systems with unstructured uncertainty

In this paper, we present novel results that parameterize a broad class of robust output-feedback model predictive control (MPC) policies for discrete-time systems with constraints and unstructured model uncertainty. The MPC policies we consider employ: (i) a linear state estimator, (ii) a pre-determined feedback gain (iii) a set of “tighter constraints” and (iv) a quadratic cost function in the degrees of freedom and the estimated state. Contained within the class, we find both well-known control policies and policies with novel features. The unifying aspect is that all MPC policies within the class satisfy a robust stability test. The robust stability test is suited to synthesis and incorporates a novel linear matrix inequality (LMI) condition which involves the parameters of the cost function. The LMI is shown to always be feasible under an appropriate small-gain condition on the pre-determined feedback gain and the state estimator. Moreover, we show, by means of both theoretical and numerical results, that choosing the cost function parameters subject to the proposed condition often leads to good nominal performance whilst at the same time guaranteeing robust stability.     

Enlargement of polytopic terminal region in NMPC by interpolation and partial invariance

This paper uses the concept of partial invariance to derive a sequence of linear programs in order to maximize offline the volume of an invariant polytopic set with an arbitrary predefined number of vertices subject to a bound on closed-loop performance. Interpolation techniques are used to determine a nonlinear control law which is optimal with respect to a closed-loop cost bound through the on-line solution of a linear program. The invariant polytope is also used to define a receding horizon control law through an appropriate terminal constraint and cost.     

Comments on: “Efficient robust constrained model predictive control with a time varying terminal constraint set” by Wan and Kothare

We present an algorithm that modifies the original formulation proposed in Wan and Kothare [Efficient robust constrained model predictive control with a time-varying terminal constraint set, Systems Control Lett. 48 (2003) 375–383]. The modified algorithm can be proved to be robustly stabilizing and preserves all the advantages of the original algorithm, thereby overcoming the limitation pointed out recently by Pluymers et al. [Min–max feedback MPC using a time-varying terminal constraint set and comments on “Efficient robust constrained model predictive control with a time-varying terminal constraint set”, Systems Control Lett. 54 (2005) 1143–1148].     

Least-restrictive move-blocking model predictive control

Move-blocking lowers the computational complexity of model predictive control (MPC) problems by reducing the number of optimization variables. However, this may render states close to constraints infeasible. Thus move-blocking generally results in control laws that are restrictive; the controller domains may be unacceptably and unnecessarily small. Furthermore, different move-blocking strategies may result in controller domains of different sizes, all other factors being equal. In this paper an approach is proposed to design move-blocking MPC control laws that are least-restrictive, i.e. the controller domain is equal to the maximum controlled invariant set. The domains of different move-blocking controllers are then by design equal to each other. This allows comparison of differing move-blocking strategies based on cost performance only, without needing to consider domain size also. Thus this paper is a step towards being able to derive optimal move-blocking MPC control laws.     

Predictive metamorphic control

In spite of its easy implementation, ability to handle constraints and nonlinearities, etc., model predictive control (MPC) does have drawbacks including tuning difficulties. In this paper, we propose a refinement to the basic MPC strategy by incorporating a tuning parameter such that one can move smoothly from an existing controller to a new MPC strategy. Each change of this tuning parameter leads to a new stabilising control law, therefore, allowing one to gradually move from an existing control law to a new and better one. For the infinite horizon case without constraints and for the general case with state and input constraints, stability results are established. We also examine the practical applicability of the proposed approach by employing it in the nominal prediction model of the tube-based output feedback robust MPC method. The merits of the proposed method are illustrated by examples.     

Generalized terminal state constraint for model predictive control

A terminal state equality constraint for Model Predictive Control (MPC) laws is investigated, where the terminal state/input pair is not fixed a priori but it is a free variable in the optimization. The approach, named “generalized” terminal state constraint, can be used for both tracking MPC (i.e. when the objective is to track a given steady state) and economic MPC (i.e. when the objective is to minimize a cost function which does not necessarily attains its minimum at a steady state). It is shown that the proposed technique provides, in general, a larger feasibility set with respect to the existing approaches, given the same prediction horizon. Moreover, a new receding horizon strategy is introduced, exploiting the generalized terminal state constraint. Under mild assumptions, the new strategy is guaranteed to converge in finite time, with arbitrarily good accuracy, to an MPC law with an optimally-chosen terminal state constraint, while still enjoying a larger feasibility set. The features of the new technique are illustrated by an inverted pendulum example in both the tracking and the economic contexts.     

On the explicit dead-time compensation for robust model predictive control

Computational simplicity is one of the most important aspects to take into account in robust model predictive control (MPC). In dead-time processes, it is common to use an augmented state-space representation in order to apply robust MPC strategies but, this procedure may affect computational aspects. In this paper, explicit dead-time compensation will be used to avoid augmented representation. This technique will be analyzed in terms of robust stability and constraint satisfaction for discrete-time linear systems. The results of this discussion will be applied to a robust tube-based MPC strategy which is able to guarantee robust stability and constraint satisfaction of a dead-time system by considering a prediction model without dead-time. Moreover, taking advantage of the proposed scheme, the robust MPC will be particularized for first-order plus dead-time models which simplifies significantly controller synthesis. The proposed dead-time compensation method will be applied to different robust MPC strategies in two case studies: (i) a simulated quadruple-tank system, and (ii) an experimental scaled laboratory heater process.     

Constrained linear system with disturbance: Convergence under disturbance feedback

This paper proposes a disturbance-based control parametrization under the Model Predictive Control framework for constrained linear discrete time systems with bounded additive disturbances. The proposed approach has the same feasible domain as that obtained from parametrization over the family of time-varying state feedback policies. In addition, the closed-loop system is stable in the sense that the state converges to a bounded set that has a characterization determined by a feedback gain.     

模型预测控制的新发展

随着模型预测控制（MPC）的广泛应用,其向智能化的方向发展成为必然,因此在近年来MPC发展的基础上,本文详细综述了预测控制包括多变量、约束、鲁棒、非线性等方面的工作,概述了预测控制与先进的控制算法的结合状况,并对存在的问题进行了探讨,从理论上分析了其智能化发展趋势和方向。     

Least-restrictive robust periodic model predictive control applied to room temperature regulation

State-feedback model predictive control (MPC) of constrained discrete-time periodic affine systems is considered. The periodic systems’ states and inputs are subject to periodically time-dependent, hard, polyhedral constraints. Disturbances are additive, bounded and subject to periodically time-dependent bounds. The objective is to design MPC laws that robustly enforce constraint satisfaction in a manner that is least-restrictive, i.e., have the largest possible domain. The proposed design method is demonstrated on a building climate control example. The proposed method is directly applicable to time-invariant MPC.     

More efficient predictive control

An approach for constrained predictive control of linear systems (or uncertain systems described by polytopic uncertainty models) is presented. The approach consists of (in general non-convex, but often convex) offline optimization, and very efficient online optimization. Two examples, one being a laboratory experiment, compare the approach to existing approaches, revealing both advantages and disadvantages.     

Constrained model predictive force control of an electrohydraulic actuator

An analytical MPC controller was designed for force control of a single-rod electrohydraulic actuator. The controller based on a difference equation uses short control horizon. The constraints on both input and output variables are taken into consideration by the controller. The mechanism of output constraints satisfaction uses output prediction and makes possible to constrain the output values many sampling instants ahead. Thus, it extends capabilities of the analytical MPC controllers to the field reserved so far for much more computationally expensive numerical MPC algorithms. Results of real life experiments illustrate efficiency of the proposed controller. The results also show that the MPC controller has better tracking performance than conventional P and PI controllers. The MPC controller with the constraint handling mechanisms, though relatively simple, offers very good performance. As the design process is detailed, it is possible to relatively easy adapt the proposed approach to other control plants.     

Controlled invariant feasibility — A general approach to enforcing strong feasibility in MPC applied to move-blocking

Strong feasibility of MPC problems is usually enforced by constraining the state at the final prediction step to a controlled invariant set. However, such terminal constraints fail to enforce strong feasibility in a rich class of MPC problems, for example when employing move-blocking. In this paper a generalized, least restrictive approach for enforcing strong feasibility of MPC problems is proposed and applied to move-blocking MPC. The approach hinges on the novel concept of controlled invariant feasibility. Instead of a terminal constraint, the state of an earlier prediction step is constrained to a controlled invariant feasible set. Controlled invariant feasibility is a generalization of controlled invariance. The convergence of well-known approaches for determining maximum controlled invariant sets, and j-step admissible sets, is formally proved. Thus an algorithm for rigorously approximating maximum controlled invariant feasible sets is developed for situations where the exact maximum cannot be determined.     

Robust model predictive control with guaranteed setpoint tracking

In this paper a novel robust model predictive control (RMPC) algorithm is proposed, which is guaranteed to stabilize any linear time-varying system in a given convex uncertainty region while respecting state and input constraints. Moreover, unlike most existing RMPC algorithms, the proposed algorithm is guaranteed to remove steady-state offset in the controlled variables for setpoints (possibly) different from the origin when the system is unknown linear time-invariant. The controller uses a dual-mode paradigm (linear control law plus free control moves to reach an appropriate invariant region), and the key step is the design of a robust linear state feedback controller with integral action and the construction of an appropriate polyhedral invariant region in which this controller is guaranteed to satisfy the process constraints. The proposed algorithm is efficient since the on-line implementation only requires one to solve a convex quadratic program with a number of decision variables that scale linearly with the control horizon. The main features of the new control algorithm are illustrated through an example of the temperature control of an open-loop unstable continuous stirred tank reactor.     

Dynamic optimisation of the aroma production in brewing fermentation

Several key compounds for the final beer flavour (higher alcohols, esters, vicinal diketones) are produced during the alcoholic fermentation phase. The paper demonstrates the possibility of obtaining various desired final aroma profiles and reducing the total process time using dynamic optimisation of three control variables: temperature, top pressure and initial yeast concentration in the fermentation tank. The optimisation is based on a sequential quadratic programming algorithm, on a dynamic model of the alcoholic fermentation and on an aroma production model. The robustness of the optimal control profile with respect to model uncertainty is discussed.