Event-triggered adaptive critic learning for the unmatched uncertain and asymmetric input-constrained systems using particle swarm optimization

Practical applications often involveunmatched uncertainties and asymmetric input constraints. This paper focuses on designing a control strategy for a class of nonlinear systems with unmatched uncertainties and asymmetric input-constraints using event-triggered adaptive critic learning and particle...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Nonlinear dynamics Jg. 113; H. 8; S. 8533 - 8553
Hauptverfasser: Chu, Zhousheng, Liu, Chong, Yue, Hongyun, Li, Weihua
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Dordrecht Springer Nature B.V 01.04.2025
Schlagworte:
Algorithms
Asymmetry
Closed loops
Constraints
Control systems
Controllers
Distance learning
Dynamic programming
Feedback control
Machine learning
Neural networks
Nonlinear systems
Optimal control
Particle swarm optimization
Uncertainty
ISSN:0924-090X, 1573-269X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Practical applications often involveunmatched uncertainties and asymmetric input constraints. This paper focuses on designing a control strategy for a class of nonlinear systems with unmatched uncertainties and asymmetric input-constraints using event-triggered adaptive critic learning and particle swarm optimization algorithm (PSOA). The unmatched uncertainties are first decomposed to construct an auxiliary control system. Through theoretical analysis, the control policy is derived from a transformed optimal control problem by defining a performance function. A neural network (NN)-based adaptive critic learning technology, known as adaptive dynamic programming (ADP) method, is used to iteratively search for the optimal controller. Unlike the gradient descent algorithm, the PSOA is used to train the NN weights, which enhances the reliability of online learning. It is further proved that all signals in the closed-loop system are uniformly ultimately bounded (UUB). Additionally, to conserve communication and computational resources, an event-triggered mechanism is incorporated to update the controller. Finally, three detailed simulation results show the effectiveness of the proposed method.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0924-090X
1573-269X
DOI:10.1007/s11071-024-10574-2