Luís C. Oliveira-Lopes

Experimental assessment and design of multiple model predictive control based on local model networks for industrial processes

This paper presents the experimental assessment of a class of multiple model predictive controllers based on linear local model networks. The control design is established on a clear and easily understandable structure where local models are used to describe the nonlinear process in several different operating points. Thus, simplifying the model predictive control (MPC) algorithm by eliminating the need of a nonlinear optimization strategy, reducing it to a well-grounded quadratic programming dynamic optimization problem. The investigated methodology was tested in an experimental neutralization pilot plant instrumented with foundation fieldbus devices. The steps of models selection, experimental system identification and proper model validation were addressed. In addition, an open-source system used for control calculation was presented. Results regarding the control problem showed that the MPC based on local relevant models was capable of smooth setpoint tracking despite system nonlinearities and a reduced demand of the final control element was attained. It was shown that the technique is easily implementable and can be used to achieve improvements in the control of nonlinear processes at the cost of little modification to the linear MPC algorithm.

CONTROL RECONFIGURATION OF CHEMICAL PROCESSES SUBJECTED TO ACTUATOR FAULTS: A MOVING HORIZON APPROACH

The use of Fault Tolerant Control (FTC) strategies provides alternatives for control loops in presence of faults by exploring physical and analytical redundancies available in the process. Therefore, the use of FTC techniques is essential to adequate industrial control systems concerning availability and reliability while preserving the closed-loop system performance. In this paper, we propose the extension of the fault-hiding approach for the control reconfiguration to a dynamic optimization problem in order to incorporate system constraints. This method uses the concepts of the virtual actuator and a moving horizon framework extending the original control reconfiguration problem to a quadratic programming problem. A multivariable neutralization process subjected to communication loss between the master controller and the final control elements is used as an example to analyze and demonstrate the performance of the proposed approach.

A Cooperative Distributed Model Predictive Control for Nonlinear Systems with Automatic Partitioning

Abstract Distributed Model Predictive Control (DMPC) algorithms in which each local controller optimizes a global cost function are known as cooperative. This work investigates a cooperative DMPC coupled to an automatic partitioning of the nonlinear system. The local decomposition used is based on the directed graph of the effects on control variables from the space of inputs through a successive linearization of the plant model at each sampling instant. The proposed strategy was applied to a benchmark plant and compared to centralized controller.

Reconfigurable stabilizing control applied to a neutralization process

Abstract This paper presents the application of reconfigurable control strategy to a multivariable neutralization process under several fault scenarios. The combined control strategy is integrated to a reconfiguration block that uses the available signals from the remaining sensors and actuators to reconstruct a signal useful to the nominal controller leading to a stable closed-loop behavior even in the presence of faults. Simulations results show that, although the overall system performance can be compromised, the reconfigurable control strategy leads to a system that is still able to operate within acceptable performance in the studied fault scenarios.

Fuzzy Model Predictive Control Applied to Piecewise Linear Systems

This paper presents the application of a combined control strategy applied to nonlinear systems and grounded in a MPC structure using a fuzzy model description. The nonlinear models were treated as several sub-models with linear behavior, called Piecewise Linear systems (PWL). The advantages of this methodology were visualized in the temperature control of a continuous tank reactor with output multiplicity behavior. Comparisons with classical control approaches showed that the PWL MPC is an attractive and practical strategy.

Hybrid System Descriptions for Chemical Engineering Processes

Abstract The hybrid description for process modeling has the advantage of allowing a more flexible description for dynamic models in the presence of state jumps and logical decisions. This work presents hybrid descriptions for processes of interest in Chemical Engineering, therefore, phenomena such as change in the number of stages/phases, processes failures, tolerant control, procedures or arrangements between stages, and jump on variables may be assessed by the verification of computational hybrid models. Cases of interest of process modeling, simulation, control and analysis are discussed.

On l1-Predictive Control of Mixed-Logical Dynamical Systems

Abstract Constant technological innovations and environmental and economical requirements have created new challenges in industrial process control. The hybrid system predictive control is a representative example of great interest in this modern scenario. Hybrid behavior is found by continuous operation dynamics inserted with discrete event changes that might occur by internal and external actions of the system. Internal actions that generate hybrid behavior can be faults, changes between operation modes and perturbations. External actions can be represented by logical and decisory aspects and safety constraints. The chemical process operation often deals with controlled transitions among operation modes due to changes in the raw materials, energy sources, product specification and market demand. In this paper, a hybrid systems representation of linear systems known as mixed logical dynamical models (MLD) is investigated. The paper addresses the MLD /1-Model Predictive Control (MPC) problem. Structurally, the MLD model control problem is described through a linear mixed integer optimization problem (MILP), which is computationally demanding, but they can be satisfactorily applied to systems of certain speed and dimension. The paper presents a study on the characteristics of the controller. A system of interest is used to illustrate the study. The results show that the application of MLD in the MPC control framework can have a positive impact specially when there is the inclusion of logical decisions aspects and operational knowledge in the MPC problem.