Evolutionary Algorithms for the Multiple Unmanned Aerial Combat Vehicles Anti-ground Attack Problem in Dynamic Environments


Abstract

This chapter aims to solve the online path planning (OPP) and dynamic target assignment problems for the multiple unmanned aerial combat vehicles (UCAVs) anti-ground attack task using evolutionary algorithms (EAs). For the OPP problem, a model predictive control framework is adopted to continuously update the environmental information for the planner. A dynamic multi-objective EA with historical Pareto set linkage and prediction is proposed to optimize in the planning horizon. In addition, Bayesian network and fuzzy logic are used to quantify the bias value to each optimization objective so as to intelligently select an executive solution from the Pareto set. For dynamic target assignment, a weapon target assignment model that considers the inner dependence among the targets and the expected damage type is built up. For solving the involved dynamic optimization problems, an environment identification based memory scheme is proposed to enhance the performance of estimation of distribution algorithms. The proposed approaches are validated via simulation with a scenario of suppression of enemy air defense mission.