Design of a Model Predictive Controller Based on Data-Driven Approach

Please enable javascript in your browser.

Mehrnoosh, Amir | 2023

37 Viewed

Type of Document: M.Sc. Thesis
Language: Farsi
Document No: 55811 (05)
University: Sharif University of Technology
Department: Electrical Engineering
Advisor(s): Haeri, Mohammad
Abstract:
In this dissertation, a data-driven model predictive control (MPC) algorithm has been developed to control unknown linear time invariant systems. This algorithm only needs past measured input-output data of the system and an upper bound on the order of the system to derive the system output to a desired value. To improve the transient response of the system, we define two filters on the performance and the effort terms of cost function of the optimization problem. Moreover, the Hankel matrices used in the optimization problem are updated online based on a multi-step event-triggered MPC scheme to deal with uncertainties and disturbances. With this method, the computational load of this algorithm is reduced and a balance is stablished between the computational cost and the closed-loop performance. The control objective is to track a given input-output reference. In this algorithm, using a virtual setpoint, it is possible to get as close to the desired value as possible if the reference point is unreachable for the system dynamics. The simulation results of the proposed algorithm for a first-order and a second-order system will show its efficiency
Keywords:
Predictive Control ; Optimization ; Data Driven Control Systems ; Event-Triggered Control ; Robust Control

No TOC