Data-Driven Fault-Tolerant Model Predictive Control with Online Adaptation Based on Dynamic Mode Decomposition

Document Type : Original Article

Authors

1 Department of Electrical and Computer Engineering, Qom University of Technology

2 Department of electrical and computer engineering, Qom University of Technology

3 Department of Electrical Engineering, Yazd University

Abstract

This paper introduces a new data-driven control method for handling faults in multi-input multi-output linear systems. In this fault-tolerant control approach, a virtual actuator based on model predictive control is designed, utilizing an adaptive model extracted from the system's dynamic modes. This extracted model is recursively updated and utilized to predict the states of the faulty system. The method considers the effect of faults in predictive controller predictions and minimizes this effect. The proposed method does not require knowledge of system equations and it is fully data-driven. Moreover, since it utilizes predictive controller for improving system performance in the presence of faults, it can impose constraints on control inputs. Furthermore, the method presented in this paper is designed based on a virtual actuator and can easily be augmented to closed-loop systems to enhance their performance in dealing with faults. Finally, the performance of the proposed control method is examined through a simulation example.

Keywords

Main Subjects

TABRIZ JOURNAL OF ELECTRICAL ENGINEERING

Articles in Press, Corrected Proof
Available Online from 28 September 2024

Files

Share

How to cite

Statistics