An Improved Firework Algorithm Based on Local Search Optimization

Aiming at the problems of the firework algorithm (FWA) that is easy to fall into the local optimum and the search efficiency and accuracy are low, a firework algorithm with local search optimization (LSFWA) is proposed. In this paper, we present two performance enhancing strategies. The first strategy is that the iterative local search is performed to find the current optimal location. The second strategy is the perturbation strategy, which increases the diversity of sparks to a certain extent by introducing a probability perturbation, then, search for sparks with better fitness near the local extremum to improve the ability to search for the global optimal location. The simulation experiment on 9 test functions shows that LSFWA displays better optimization speed and optimization accuracy compared to the typical algorithm FWA, FAGSO, PSO and GSA.