G. Valencia-Palomo

Programmable logic controller implementation of an auto-tuned predictive control based on minimal plant information.

This paper makes two key contributions. First, it tackles the issue of the availability of constrained predictive control for low-level control loops. Hence, it describes how the constrained control algorithm is embedded in an industrial programmable logic controller (PLC) using the IEC 61131-3 programming standard. Second, there is a definition and implementation of a novel auto-tuned predictive controller; the key novelty is that the modelling is based on relatively crude but pragmatic plant information. Laboratory experiment tests were carried out in two bench-scale laboratory systems to prove the effectiveness of the combined algorithm and hardware solution. For completeness, the results are compared with a commercial proportional-integral-derivative (PID) controller (also embedded in the PLC) using the most up to date auto-tuning rules.

Efficient algorithms for trading off feasibility and performance in predictive control

This article introduces two simple modifications to conventional predictive control algorithms with the aim of enlarging the feasible regions without increasing computational complexity. It is shown that despite the relatively large feasibility gains, the loss in performance may be far smaller than expected and thus the algorithms give mechanisms for achieving low computational loads with good feasibility and good performance while using a simple algorithm set-up. Both algorithms have standard convergence and feasibility guarantees.

Alternative parameterisations for predictive control: How and why?

This paper looks at the efficiency of the parameterisation of the degrees of freedom within an optimal predictive control algorithm. It is shown that the conventional approach of directly determining each individual future control move is not efficient in general, and can give poor feasibility when the number of degrees of freedom are limited. Two systematic alternatives are explored and both shown to be far more efficient in general.

Exploiting Kautz functions to improve feasibility in MPC

Abstract This paper develops the recently published Laguerre MPC by proposing an alternative parametrization of the degrees of freedom in order to further increase the feasible region of model predictive control (MPC). Specifically, a simple but efficient algorithm that uses Kautz functions to parameterize the degrees of freedom in Optimal MPC is presented. It is shown that this modification gives mechanisms to achieve low computation burden with good feasibility and good performance. The improvements, with respect to an existing algorithm that uses a similar strategy, are demonstrated by examples.

Using Laguerre functions to improve efficiency of multi-parametric predictive control

Multi-parametric quadratic programming (mp-QP) is an alternative means of implementing conventional predictive control algorithms whereby one transfers much of the computational load to offline calculations. However, coding and implementation of this solution may be more burdensome than simply solving the original QP. This paper shows how Laguerre functions can be used in conjunction with mp-QP to achieve a large decrease in both the online computations and data storage requirements while increasing the feasible region of the optimization problem. Extensive simulation results are given to back this claim.

Improving the feed-forward compensator in predictive control for setpoint tracking.

Simple predictive control (MPC) algorithms produce a feed-forward compensator that may be a suboptimal choice. This paper gives some insights into this issue and simple means of modifying the feed-forward to produce a more systematic and optimal design. In particular, it is shown that the optimum procedure depends upon the underlying loop tuning and also that there are, as yet under utilised, potential benefits with regard to constraint handling procedures, which helps to improve the computational efficiency of the online controller implementation. A laboratory test in a programmable logic controller (PLC) was carried out to demonstrate the code on real hardware and the effectiveness of the solution.

Auto-tuned Predictive Control Based on Minimal Plant Information

Abstract Abstract This paper makes two key contributions. First there is a definition and implementation of a novel auto-tuned predictive controller. The key novelty is that the modelling is based on relatively crude but pragmatic plant information. Secondly, the paper tackles the issue of availability of predictive control for low level control loops. Hence the paper describes how the controller is embedded in an industrial Programmable Logic Controller (PLC) using the IEC 1131.1 programming standard. Laboratory experiment tests were carried out in two bench-scale laboratory systems to prove the effectiveness of the combined algorithm and hardware solution. For completeness, the results are compared with a commercial PID controller (also embedded in the PLC) using the most up to date auto-tuning rules.

A move-blocking strategy to improve tracking in predictive control

One of the advantages of model predictive control is its ability to take optimal account of information about future setpoint changes in the design of the control law. However, it is easy to show that the default optimum GPC (Generalized Predictive Control) law that uses this information can lead to a deterioration rather than an improvement in performance. This paper discusses some simple modifications to overcome this problem. Specifically the focus here is on move-blocking approaches which seek to extend the effective input horizon while restricting the number of optimisation variables. A simple algorithm is proposed which allows optimal nominal performance and tracking without the undesirable effects arising from the standard approaches; hence the proposed modification always improves algorithm performance.

Efficient suboptimal parametric implementations for predictive control

Abstract This paper develops parametric approaches to predictive control but differs from more conventional approaches in that it pre-defines the complexity of the solution rather than the allowable suboptimality. The paper proposes a novel parameterisation of the parametric regions which allows efficiency of definition, effective spanning of feasible region and also highly efficient search algorithms. Despite the suboptimality, the algorithm retains guaranteed stability, in the nominal case.