A Comparative Study of Response of Control Problems to Unconstrained & Constrained Model Predictive Control

Abstract

Model Predictive Control which was originally developed to meet the control needs of petroleum refineries and power plants can now be found in a wide variety of application areas such as chemicals, food processing, automotive and aerospace applications. The reason for increase in use of MPC technology is its ability to handle multivariable control problems and constraints. The saturation on the control signal can make the control system performance deteriorate significantly. The strength of the constrained control framework using model predictive control lies in the optimality it achieves in a systematic manner, and the generality to cope with the multi-input, multi-output system with various constraints This dissertation deals with the Model Predictive Control, MPC, with the goal of comparing the response of control problems to constrained and unconstrained MPC. The control problem of Van de Vusse Reactor and a temperature control problem are considered to which unconstrained and constrained MPC is applied respectively. Further, a positioning platform problem is taken and its response to both unconstrained and constrained MPC is compared. In MPC strategy, a sequence of control actions (from present time onwards) is computed that minimizes a finite duration objective function that starts from the present time and extends a fixed number of time steps into the future. Although a sequence of present and future control actions is computed, only the present control action is applied to the system. At the next time step the entire process repeats. Thus, the starting and ending time steps of the objective function shifts one step forward, thus the term receding-horizon control strategy is often associated with this strategy.