Miguel Peña

Hybrid Systems for Solving Model Predictive Control of Piecewise Affine System

Abstract This paper presents a hybrid procedure to solve Model Predictive Control (MPC) of Piecewice Affine (PWA) system. The procedure uses the concepts of reacheble set, controllable set and State Transition Graph (STG) in order to reduce the number of Quadratic Problems (QP) needed to obtain an global minimum. The proposed algorithm reduces considerably the number of explorations needed during the search of a global minimum and thus the time required by the MPC can be reduced to a small fraction of the time required to the original problem

Model predictive controller for piecewise affine system

Abstract This paper presents a hybrid procedure to solve Model Predictive Controller (MPC) for Piecewise Affine System (PWA) The approach presented here belong to the class of Branch and Bound (B&B) methods. The procedure uses the concepts of reachable set combined to the specific B&B methods, in order to reduce the number of Quadratic Problems (QP) needed to be solved by the optimization algorithm.

Systems Identification using a Type of Takagi-Sugeno Fuzzy Model

Abstract A methodology for obtaining the stnlcture and parallleters of a Takagi-Sugeno fuzzy model is proposed. The algorithm is divided into two steps: coarse tuning and fine tuning. In coarse tuning, the model structure identification is based on partial derivatives obtained from the sampled output with regard to the inputs. In fme tuning, a traditional nonlinear optinlization method is used to adjust the antecedent and consequent parameter. Finally, some examples are given to demonstrate the validity of these algorithms.

Robust Predictive Control for a Greenhouse Using Input/Output Linearization and Linear Matrix Inequalities

Abstract The present work proposes a scheme for temperature control of a greenhouse. An approach based on a combination of two different control schemes: Input/Output Linearization an L.lnear Matnx InequalItIes based predictive controller design is proposed. The algorithm consists of three steps, first a feedback linearization, secondly the derivation of an uncertainty descnption and finally the LMI-based optimization. Several simulations shows the effect of the combInation of Its techniques in the performance of the inside temperature for a greenhouse.

Deterministic robust control design for an iron core linear drive including second-order electrical dynamics

This work proposes a deterministic robust controller to improve tracking performance for a linear motor, taking into account the electrical dynamics imposed by a commercial current controller. The design is split in two parts by means of the backstepping technique, in which the first part corresponds to a typical deterministic robust controller, neglecting the electrical dynamics. In the second part, a second-order electrical dynamics is considered using a particular state transformation. There, the proposed control law is composed of a term to compensate the known part of the model and a robust control term to impose a bound on the effect of uncertainties on tracking error. Stability and boundedness results for the complete controller are given. To this effect, a general result on boundedness and stability of nonlinear systems with conditionally bounded state variables is derived first. Finally, experimental results for the complete controller show an improvement on tracking error of up to 31.7% when compared with the results from the typical controller that neglects the electrical dynamics.

Fuzzy Model Predictive Control: Simultaneous and Sequential Comparison

Abstract This paper presents the application of a combined control strategy grounded in a basic Model Predictive Control (MPC) structure, but using a Discrete Fuzzy Model (DFM) of the process in the state space domain. Two methods of optimization are tested. The first one is the traditional sequential method where the model is solved at each iteration of the optimization routine. The second is the simultaneous method, which integrate the solution of the model into the optimization problem, provided the fulfillment of some conditions in the DFM. A particular DFM that fulfill such conditions is presented. The advantages and disadvantages of the two methods are exposed in accordance with the results of simulation tests. Finally, some conclusions and future research topics are given

Nonlinear Model Predictive Control via Feedback Linearization of a Greenhouse

Abstract A control scheme for nonlinear systems is proposed here, combining the techniques of Feedback Linearization (FL) and Model Predictive Control (MPC). Since the methodology used for solving the MPC+FL generally leads to an optimization problem subjected to state-dependent non-linear constraints, an alternative for implementation is discussed. Such an alternative considers an Extended Kalman Filter (EKF) to evaluate the non-measurable state variables of the greenhouse. The performance of two control techniques are compared, namely, MPC+FL and NLMPC.