Conventional and Explicit Approaches for Simultaneous Moving Horizon Estimation and Model Predictive Control: A Comparative Evaluation

Abstract

Model Predictive Control (MPC) has been applied across a wide range of engineering applications including process industries. MPC requires complete knowledge of states at the current instant which can either be measured directly or estimated using a state estimator. Of late, Moving Horizon Estimation (MHE) has been widely used as a state estimator owing to its ability to handle constraints. Both MPC and MHE involve solving an optimization problem at each sampling instant which can prove computationally burdensome for fast systems. Multi-parametric programming based explicit approaches have been proposed in literature as a possible approach for solving the online optimization problems in a computationally efficient manner. In the current work, feasibility of the explicit approaches simultaneously for both MPC and MHE is investigated using simulation as well as experimental studies on a quadruple tank setup. The computational efforts required for this simultaneous implementation of the explicit approaches for MPC and MHE are compared with the conventional optimization approach. Results indicate feasibility of multi-parametric implementation for lower horizon lengths.