A comparative study of state-of-the-art metaheuristics for solving many-objective optimization problems of fixed wing unmanned aerial vehicle conceptual design

Abstract

The complexity of aircraft design problems increases with many objectives and diverse constraints, thus necessitating effective optimization techniques. In recent years many new metaheuristics have been developed, but their implementation in the design of the aircraft is limited. In this study, the effectiveness of twelve new algorithms for solving unmanned aerial vehicle design issues is compared. The optimizers included Differential evolution for multi-objective optimization, Many-objective nondominated sorting genetic algorithm, Knee point-driven evolutionary algorithm for many-objective optimization, Reference vector guided evolutionary algorithm, Multi-objective bat algorithm with nondominated sorting, multi-objective flower pollination algorithm, Multi-objective cuckoo search algorithm, Multi-objective multi-verse optimizer, Multi-objective slime mould algorithm, Multi-objective jellyfish search algorithm, Multi-objective evolutionary algorithm based on decomposition and Self-adaptive many-objective meta-heuristic based on decomposition. The design problems include four many-objective conceptual designs of UAV viz. Conventional, Conventional with winglet, Twin boom and Canard, which are solved by all the optimizers employed. Widely used Hypervolume and Inverted Generational Distance metrics are considered to evaluate and compare the performance of examined algorithms. Friedman's rank test based statistical examination manifests the dominance of the DEMO optimization technique over other compared techniques and exhibits its effectiveness in solving aircraft conceptual design problems. The findings of this work assist in not only solving aircraft design problems but also facilitating the development of unique algorithms for such challenging issues.