Multiple model unfalsified adaptive generalized predictive control based on the quadratic inverse optimal control concept

• Published in: Optimal control applications & methods Vol. 42; no. 3; pp. 769 - 785
• Main Authors: Sadeghi Forouz, Bahman, Nouri Manzar, Mojtaba, Khaki‐Sedigh, Ali
• Format: Journal Article
• Language: English
• Published: Glasgow Wiley Subscription Services, Inc 01-05-2021
• Subjects: Actuators; Adaptive control; constrained predictive control; Constraints; Control methods; Control stability; Controllers; inverse optimal control; multi‐model unfalsified adaptive control; Optimal control; Predictive control; Signal generation; Stability analysis; switching control; Systems analysis; Transient performance
• ISSN: 0143-2087; 1099-1514
• DOI: 10.1002/oca.2700

Unfalsified adaptive control (UAC) is a class of switching control systems which deals with the control of uncertain systems. The UAC includes a bank of controllers, a supervisor, and a system in which the supervisor selects a stabilizing controller based on the system input and output data. Feasibility is the only assumption required in the UAC strategy, which guarantees that there is at least one stabilizing controller in the controller bank. UAC uses the cost detectability definition to prove closed‐loop stability. The combination of UAC and multiple model supervisory adaptive control (MMASC) results in the proposed unfalsified multi‐model control methodology that enjoys appropriate transient performance and stability proof with required minimum assumptions. In practical controller implementations, the effect of actuator constraints on the control signals is crucial. Despite the significance of constrained systems analysis in real applications, the input constraints in the structures of unfalsified control are not generally considered. Also, the stability analysis of constrained unfalsified control is key to its practical applications. In this article, the input constrained systems are considered using the constrained generalized predictive control (GPC) as the main controllers. Subsequently, to handle virtual signal generation for the GPCs in the UAC context, the inverse optimal control strategy is engaged and formulated to solve the signal generation problem. Simulation results are employed to show the effectiveness of the proposed methodology.