摘要：To solve the problems of the waste of resources caused by the small explosion amplitude (even close to 0) of the best firework in the conventional fireworks algorithm (FWA), and the relatively weak local search ability caused by the minimal explosion amplitude check method in the enhanced fireworks algorithm (EFWA), this paper proposes two improved strategies for FWA. First, based on the heuristic information of the distance between the best firework and other fireworks, an adaptive explosion amplitude strategy is proposed to search the local area accurately at the final phase of the FWA. Second, the highly random Lévy flight strategy is adopted instead of the Gaussian sparks strategy to generate mutation sparks in the EFWA to enhance the diversity of local search. Simulation results on 12 standard benchmark functions and their shifted functions indicate that the proposed algorithm improves the optimization precision and obtains better performance in high-dimensional complex optimization problems compared with the EFWA, the hybrid fireworks algorithm with differential evolution operator (FWADE), and the particle swarm optimization algorithm (PSO).
