SOLVING THE WEAPON–TARGET ASSIGNMENT PROBLEM BASED ON HEURISTIC ALGORITHMS

Abstract

Based on a synthesis of existing heuristic algorithms for the weapon–target assignment (WTA) problem [1], [2], this paper proposes algorithmic improvements and develops a simulation software toolkit to test and evaluate the operational performance of the enhanced algorithms. Specifically, this paper presents the development and application of the following algorithms to solve the WTA problem: an Advanced Maximum Marginal Return (Advanced MMR) algorithm, an Improved Genetic Algorithm (IGA), and an Improved Discrete Particle Swarm Optimization (IDPSO) algorithm. In particular, the proposed IDPSO incorporates an MMR-based heuristic initialization mechanism and an internal local refinement process. An offline Bayesian optimization-based hyperparameter tuning procedure [3] is used to reduce the dependence on manually selected algorithmic parameters. Experimental results obtained from the C++ simulation toolkit based on Microsoft Foundation Classes (MFC) show that, under identical input conditions and predefined time budgets, the IDPSO achieves lower expected objective-function values and higher stability than Advanced MMR and IGA in time-constrained command-and-control simulation settings.