Local consensus based multi-objective distributed optimization and its application

In multi-agent networks, the objectives of the agents are often in conflict with each other. Most existing distributed optimization models optimize a single performance metric, which is usually a weighted sum of individual objectives. Thus it cannot fully reflect the trade-off among the objectives....

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Systems & control letters Ročník 207; s. 106290
Hlavní autoři: Guo, Jieyuan, Shao, Lizhen, Lv, Quanxiu, Liang, Shu
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier B.V 01.01.2026
Témata:
Local consensus constraint
Multi-objective distributed optimization
Pareto frontier
ɛ-constraint method
Local consensus constraint
ɛ-constraint method
Pareto frontier
Multi-objective distributed optimization
ISSN:0167-6911
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In multi-agent networks, the objectives of the agents are often in conflict with each other. Most existing distributed optimization models optimize a single performance metric, which is usually a weighted sum of individual objectives. Thus it cannot fully reflect the trade-off among the objectives. In this paper, multi-objective distributed convex optimization problems with local optimal consensus in multi-agent systems are studied. A multi-objective distributed optimization algorithm based on the ɛ-constraint method is proposed and the convergence of the algorithm is proved. By changing ɛ value, the algorithm provides the decision maker with a set of representative efficient solutions to aid decision analysis. The effectiveness of the proposed method is verified through two numerical simulation examples. Furthermore, the proposed algorithm is applied to the classification problem with distributed data storage. The experimental results show that the proposed algorithm can effectively solve large-scale multi-objective optimization problems with distributed data storage. •A multi-objective distributed ɛ-constraint algorithm is proposed.•Convergence is proved and effectiveness is shown for convex multi-objective problems.•The algorithm is applied to classification with distributed data storage across agents.•UCI dataset results demonstrate the algorithm’s practicality.
ISSN:0167-6911
DOI:10.1016/j.sysconle.2025.106290