Improved dynamic programming method for solving multi-objective and multi-stage decision-making problems

Multi-objective and multi-stage decision-making problems require balancing multiple objectives at each stage and making optimal decision in multi-dimensional control variables, where the commonly used intelligent optimization algorithms suffer from low solving efficiency. To this end, this paper pro...

Full description

Saved in:
Bibliographic Details
Published in:Scientific reports Vol. 15; no. 1; pp. 1668 - 14
Main Authors: Liang, Zhihao, Zhao, Kegang, He, Kunyang, Liu, Yanwei
Format: Journal Article
Language:English
Published: London Nature Publishing Group UK 11.01.2025
Nature Publishing Group
Nature Portfolio
Subjects:
639/705/1041
639/705/117
Algorithms
Decision making
Dynamic programming
Efficiency
Elitism
Genetic algorithms
Humanities and Social Sciences
Improved dynamic programming method
Methods
Multi-objective and multi-stage decision-making problems
multidisciplinary
Non-dominated sorting
Objectives
Optimization algorithms
Pareto optimum
Performance management
Science
Science (multidisciplinary)
Solving efficiency
Improved dynamic programming method
Solving efficiency
Multi-objective and multi-stage decision-making problems
Non-dominated sorting
ISSN:2045-2322, 2045-2322
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Multi-objective and multi-stage decision-making problems require balancing multiple objectives at each stage and making optimal decision in multi-dimensional control variables, where the commonly used intelligent optimization algorithms suffer from low solving efficiency. To this end, this paper proposes an efficient algorithm named non-dominated sorting dynamic programming (NSDP), which incorporates non-dominated sorting into the traditional dynamic programming method. To improve the solving efficiency and solution diversity, two fast non-dominated sorting methods and a dynamic-crowding-distance based elitism strategy are integrated into the NSDP algorithm. The proposed algorithm has been verified on 12 benchmark test functions and a multi-objective travelling salesman problem. The results demonstrate that the NSDP algorithm achieves better outcome in multiple performance metrics and higher solving efficiency, compared with non-dominated sorting genetic algorithm II and multi-objective particle swarm optimization.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:2045-2322
2045-2322
DOI:10.1038/s41598-024-83037-8