This paper considers global optimization of a continuous nonconvex piecewise affine (PWA) function over a polytope. This type of optimization problem often arises in the context of control of continuous PWA systems. In literature, it has been shown that the given problem can be formulated as a mixed integer linear programming (MILP) problem, the worst-case complexity of which grows exponentially with the number of polyhedral subregions in the domain of the PWA function. In this paper, we propose a solution approach that is more efficient for continuous PWA functions with a large number of polyhedral subregions. The proposed approach is based on optimistic optimization, which uses hierarchical partitioning of the feasible set and which can guarantee bounds on the suboptimality of the returned solution with respect to the global optimum given a prespecified finite number of iterations. Since the function domain is a polytope with arbitrary shape, we introduce a partitioning approach by employing Delaunay triangulation and edgewise subdivision. Moreover, we derive the analytic expressions for the core parameters required by optimistic optimization for continuous PWA functions. The numerical example shows that the resulting algorithm is faster than MILP solvers for PWA functions with a large number of polyhedral subregions.

@en

This article discusses two issues in connection with the article J. Xu, T. J. J. van den Boom, and B. De Schutter, 'Model predictive control for stochastic max-plus linear systems with chance constraints,' IEEE Trans. Autom. Control, vol. 64, no. 1, pp. 337-342, Jan. 2019, namely an error in Proposition 7 and the assumption that the covariance matrix in the chance constraint is positive definite. First, we will discuss and correct the error in Proposition 7. Subsequently, we will consider a relaxation of the assumption in Proposition 7 and give a less-restrictive and less-conservative reformulation of the proposition.@en

The objective of this paper is to provide a concise introduction to the max-plus algebra and to max-plus linear discrete-event systems. We present the basic concepts of the max-plus algebra and explain how it can be used to model a specific class of discrete-event systems with synchronization but no concurrency. Such systems are called max-plus linear discrete-event systems because they can be described by a model that is “linear” in the max-plus algebra. We discuss some key properties of the max-plus algebra and indicate how these properties can be used to analyze the behavior of max-plus linear discrete-event systems. Next, some control approaches for max-plus linear discrete-event systems, including residuation-based control and model predictive control, are presented briefly. Finally, we discuss some extensions of the max-plus algebra and of max-plus linear systems.

@en

The topic of this paper is model predictive control (MPC) for max-plus linear systems with stochastic uncertainties the distribution of which is supposed to be known. We consider linear constraints on the inputs and the outputs. Due to the uncertainties, these linear constraints are formulated as probabilistic or chance constraints, i.e., the constraints are required to be satisfied with a predefined probability level. The proposed chance constraints can be equivalently rewritten into a max-affine (i.e., the maximum of affine terms) form if the linear constraints are monotonically nondecreasing as a function of the outputs. Based on the resulting max-affine form, two methods are developed for solving the chance-constrained MPC problem for stochastic max-plus linear systems. Method 1 uses Boole's inequality to convert the multivariate chance constraint into univariate chance constraints for which the probability can be computed more efficiently. Method 2 employs Chebyshev's inequality and transforms the chance constraint into linear constraints on the inputs. The simulation results for a production system example show that the two proposed methods are faster than the Monte Carlo simulation method and yield lower closed-loop costs than the nominal MPC method.

@en

This PhD thesis considers the development of optimization and model-based control techniques for max-plus linear (MPL) and continuous piecewise affine (PWA) systems. The three main topics investigated in this thesis are as follows: 1. Optimistic optimization and planning for model-based control of MPL systems; 2. Optimistic optimization for MPC of continuous PWA systems; 3. MPC for stochastic MPL systems with chance constraints@en

We consider the infinite-horizon optimal control of discrete-time, Lipschitz continuous piecewise affine systems with a single input. Stage costs are discounted, bounded, and use a 1 or ∞-norm. Rather than using the usual fixed-horizon approach from model-predictive control, we tailor an adaptive-horizon method called optimistic planning for continuous actions (OPC) to solve the piecewise affine control problem in receding horizon. The main advantage is the ability to solve problems requiring arbitrarily long horizons. Furthermore, we introduce a novel extension that provides guarantees on the closed-loop performance, by reusing data (“learning”) across different steps. This extension is general and works for a large class of nonlinear dynamics. In experiments with piecewise affine systems, OPC improves performance compared to a fixed-horizon approach, while the data-reuse approach yields further improvements.

@en

This paper considers model predictive control for continuous piecewise affine (PWA) systems. In general, this leads to a nonlinear, nonconvex optimization problem. We introduce an approach based on optimistic optimization to solve the resulting optimization problem. Optimistic optimization is based on recursive partitioning of the feasible set and is characterized by an efficient exploration strategy seeking for the optimal solution. The advantage of optimistic optimization is that one can guarantee bounds on the suboptimality with respect to the global optimum for a given computational budget. The 1-norm and ∞-norm objective functions often considered in model predictive control for continuous PWA systems are continuous PWA functions. We derive expressions for the core parameters required by optimistic optimization for the resulting optimization problem. By applying optimistic optimization, a sequence of control inputs is designed satisfying linear constraints. A bound on the suboptimality of the returned solution is also discussed. The performance of the proposed approach is illustrated with a case study on adaptive cruise control.@en

Model predictive control for max-plus linear discrete-event systems usually leads to a nonsmooth nonconvex optimization problem with real valued variables, which may be hard to solve efficiently. An alternative approach is to transform the given problem into a mixed integer linear programming problem. However, the computational complexity of current mixed integer linear programming algorithms increases in the worst case exponentially as a function of the prediction horizon. The focus of this paper is on making optimistic optimization suited to solve the given problem. Optimistic optimization is a class of algorithms that can find an approximation of the global optimum for general nonlinear optimization. A key advantage of optimistic optimization is that one can specify the computational budget in advance and guarantee bounds on the suboptimality with respect to the global optimum. We prove that optimistic optimization can be applied for the given problem by developing a dedicated semi-metric and by proving it satisfies the necessary requirements for optimistic optimization. Moreover, we show that the complexity of optimistic optimization is exponential in the control horizon instead of the prediction horizon. Hence, using optimistic optimization is more efficient when the control horizon is small and the prediction horizon is large.@en

This paper addresses the infinite-horizon optimal control problem for max-plus linear systems where the considered objective function is a sum of discounted stage costs over an infinite horizon. The minimization problem of the cost function is equivalently transformed into a maximization problem of a reward function. The resulting optimal control problem is solved based on an optimistic planning algorithm. The control variables are the increments of system inputs and the action space is discretized as a finite set. Given a finite computational budget, a control sequence is returned by the optimistic planning algorithm. The first control action or a subsequence of the returned control sequence is applied to the system and then a receding-horizon scheme is adopted. The proposed optimistic planning approach allows us to limit the computational budget and also yields a characterization of the level of near-optimality of the resulting solution. The effectiveness of the approach is illustrated with a numerical example. The results show that the optimistic planning approach results in a lower tracking error compared with a finite-horizon approach when a subsequence of the returned control sequence is applied.@en