Adaptive immune genetic algorithm for weapon system portfolio optimization in military big data environment

Military applications are producing massive amounts of data due to the use of multiple types of sensors on the battlefield. The aim of this paper is to investigate the weapon system portfolio problem with the valuable knowledge extracted from these sensor data. The objective of weapon system portfolio optimization is to determine the appropriate assignment of various weapon units, which maximizes the expected damage of all hostile targets, while satisfying a set of constraints. This paper presents a mixed integer non-linear optimization model for the weapon system portfolio problem. In order to solve this model, an adaptive immune genetic algorithm using crossover and mutation probabilities that are automatically tuned in each generation is proposed. A ground-based air defensive scenario is introduced to illustrate the feasibility and efficiency of our proposed algorithm. In addition, several large-scale instances that are produced by a test-case generator are also considered to demonstrate the scalability of the algorithm. Comparative experiments have shown that our algorithm outperforms its competitors in terms of convergence speed and solution quality, and it is competent for solving weapon system portfolio problems under different scales.