Title: Identification for control approach to data-driven model predictive control

Authors: Yadollah Zakeri; Farid Sheikholeslam; Mohammad Haeri

Addresses: Department of Electrical and Computer Engineering, Isfahan University of Technology, Isfahan, 84156-83111, Iran ' Department of Electrical and Computer Engineering, Isfahan University of Technology, Isfahan, 84156-83111, Iran ' Department of Electrical Engineering, Sharif University of Technology, Tehran, 11155-4363, Iran

Abstract: Data-driven model predictive control has become an attractive research subject in recent years. There are two main approaches to design data-driven model predictive control; direct methods in which controller is identified directly, and indirect ones where controller is designed based on an identified model of the plant. In this paper a new method for direct and offline identification of predictive controller is developed. Here, data-driven MPC design problem is reduced to basic model reference data-driven control problem which is more recognised, and can be analysed and designed by existing methods for LTI systems. The required tuning methods are adapted and four algorithms using modified correlation-based and virtual reference feedback tunings are developed. In comparison with other works, the proposed method can be used for both performance-oriented and model reference criteria, is direct, offline, and identifiable. Meanwhile, stability, and feasibility of the proposed algorithms can be guaranteed and certified using established analysis and synthesis methods for data-driven control.

Keywords: data-driven control; direct data-driven control; data-driven model predictive control; identification for control.

DOI: 10.1504/IJAAC.2024.138244

International Journal of Automation and Control, 2024 Vol.18 No.3, pp.281 - 301

Received: 21 Sep 2022
Accepted: 25 Apr 2023
Published online: 30 Apr 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article