.png){kind=icon}
.png){kind=icon}
Return
Constraint-Feature-Guided Evolutionary Algorithms for Multi-Objective Multi-Stage Weapon-Target Assignment Problems
.png){kind=icon}
PRE
AI
DOI：10.1007/s11424-025-4232-2.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The allocation of heterogeneous battlefield resources is crucial in Command and Control (C2). Balancing multiple competing objectives under complex constraints so as to provide decision-makers with diverse feasible candidate decision schemes remains an urgent challenge. Based on these requirements, a constrained multi-objective multi-stage weapon-target assignment (CMOMWTA) model is established in this paper. To solve this problem, three constraint-feature-guided multi-objective evolutionary algorithms (CFG-MOEAs) are proposed under three typical multi-objective evolutionary frameworks (i.e., NSGA-II, NSGA-III, and MOEA/D) to obtain various high-quality candidate decision schemes. Firstly, a constraint-feature-guided reproduction strategy incorporating crossover, mutation, and repair is developed to handle complex constraints. It extracts common row and column features from different linear constraints to generate the feasible offspring population. Then, a variable-length integer encoding method is adopted to concisely denote the decision schemes. Moreover, a hybrid initialization method incorporating both heuristic methods and random sampling is designed to better guide the population. Systemic experiments are conducted on three CFG-MOEAs to verify their effectiveness. The superior algorithm CFG-NSGA-II among three CFG-MOEAs is compared with two state-of-the-art CMOMWTA algorithms, and extensive experimental results demonstrate the effectiveness and superiority of CFG-NSGA-II.
Keywords:
Evolutionary algorithms
constrained multi-objective optimization problem
constraint handling
weapon-target assignment
Journal
IF:
2.8
Papers:
212
Citations:
2.1K