Modeling and Predictive Control of Nonlinear Hybrid Systems Using Mixed Logical Dynamical Formalism

Abstract

This work deals with the modeling and the control of hybrid systems by using Mixed Logical Dynamical (MLD) system framework described by interdependent physical laws, logic rules, and operating constraints. These are describe by linear dynamic equations subject to linear inequalities involving real and integer variables. The changes which may appear over such dynamics, are modeled by using the auxiliary variables which take into account the interconnections effects. The MLD model is used to synthesize the model predictive control law (MPC). The discrete-time equivalent of the model predicts the hybrid system behavior over a prediction horizon. The controller requires solution of on line mixed integer quadratic or linear program to solve an optimization problem. Simulation was performed using HYSDEL compiler and APROS software to illustrate performances and efficiently of these tools using the model of a three-tank COSY benchmark.