Doctoral Thesis Defense Session No. 56 of PPGEEC - Model-Based Predictive Control Strategies for Dynamic Target Reference Tracking

STUDENT: BRUNO SCHETTINI SOARES PEREIRA

DATE: 12/19/2024

TIME: 2:00 PM

LOCATION: https://conferenciaweb.rnp.br/ufba/defesas-ppgee

TITLE: Model-Based Predictive Control Strategies for Dynamic Target Reference Tracking

KEYWORDS: Model-Based Predictive Control; Constrained Systems; Trajectory Tracking;

ABSTRACT: Model-based predictive controllers (MPC) are widely used to control systems with characteristics common to many industrial and academic applications. The control law of an MPC is formulated by minimizing an objective function that considers the error between estimated predicted outputs and future references over a specified time horizon. These predicted outputs can be defined using a nominal parametric model or a data-driven model. Typically, in linear approaches, the minimization process generates an optimal sequence of inputs through the sliding horizon principle, naturally incorporating the system constraints in its solution. A known problem with constrained MPC is the loss of feasibility when there is a change of references. This happens due to the finite horizon chosen and the constraints imposed by the controller that can lead the optimizer to an unattainable target, depending on the initial conditions of the problem, which can cause reference tracking failure. This work aims to present contributions to MPC strategies when subject to time-varying references. It will be seen that the MPC presents a conflicting cost function in such conditions, so that a simple modification can provide flexibility to the MPC tuning, thus avoiding unwanted problems of high controller gains caused by aggressive tuning. Initially, a filtered DMC approach is evaluated, in which the prediction filter maintains sensitivity to high-frequency disturbances, while the MPC tuning improves reference tracking, and its effectiveness is evaluated in simulation and experiment of the temperature control of a thermoresistive sensor. Then, a modified cost function approach in a receptance-based MPC was addressed in order to reduce the nominal value of the cost function under time-varying reference conditions. The receptance matrix-based modeling proposes to bring simplicity in the identification step of complex systems, in such a way that parameters of the phenomenological model of a multi-body system do not necessarily need to be known. Thus, simulation results were evaluated for an underactuated system, showing its effectiveness for tracking periodic references. The MPC problem with robust stability assurance will also be addressed in this work through the use of artificial references with analytical reformulation of the desired target, in which its efficiency is exposed by numerical simulation analysis and its practical applicability is demonstrated through a practical application of the trajectory control of a UGV.

MEMBERS OF THE BOARD:

TITO LUIS MAIA SANTOS (ADVISOR)                  UFBA

TIAGO TRINDADE RIBEIRO                                UFBA

ACBAL RUCAS ANDRADE ACHY                          UFRB

ANTONO DA SILVA SILVEIRA                             UFPA

BISMARK CLAURE TORRICO                              UFC   

Em 06/12/2024