Momentum-Based Multiagent Approaches to Distributed Nonconvex Optimization

In this article, a paradigm of momentum-based systems is introduced for nonconvex optimization. Based on the paradigm, a momentum-based system and a momentum-based multiagent system are developed for nonconvex constrained optimization and distributed nonconvex optimization, respectively, and the con...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:IEEE transactions on automatic control Ročník 70; číslo 5; s. 3331 - 3338
Hlavní autori: Xia, Zicong, Liu, Yang, Kou, Kit Ian, Lu, Jianquan, Gui, Weihua
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: New York IEEE 01.05.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Predmet:
Algorithms
Convergence
Convex functions
Distributed nonconvex optimization
Heuristic methods
Linear programming
Momentum
momentum-based optimization
Multiagent systems
multiagent systems (MASs)
Neurodynamics
Optimization
Particle swarm optimization
Power system stability
Recurrent neural networks
Scalability
Security
Swarm intelligence
ISSN:0018-9286, 1558-2523
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this article, a paradigm of momentum-based systems is introduced for nonconvex optimization. Based on the paradigm, a momentum-based system and a momentum-based multiagent system are developed for nonconvex constrained optimization and distributed nonconvex optimization, respectively, and the convergence and convergence rate to a local optimal solution are proven. In addition, a hybrid swarm intelligence algorithm is established, which consists of multiple momentum-based systems for scattering searches and a meta-heuristic rule for repositioning the states upon their local convergence. Two numerical examples are elaborated to verify and demonstrate the optimality, enhanced stability, and faster convergence of the proposed approaches.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9286
1558-2523
DOI:10.1109/TAC.2024.3522188