An Improved Fruit Fly Optimization Algorithm for Solving High Dimensional Problems

Abstract

As a new optimization algorithm, Fruit Fly Optimization Algorithm (FOA) attracts a lot of attentions. By analyzing the probability of FOA jumping out of the local optimal range, we verified that FOA is ineffective in solving complex optimization problems whose optimal solution is nonzero. In order to improve the performance of FOA, a Modified Global Fruit Fly Optimization Algorithm (MGFOA) is introduced in this paper. In MGFOA, a uniform mechanism to produce the candidate solution is used to improve the global searching ability, a self-adaptive way to control the flight range is adapted to increase the optimize accuracy, and a ladder growth way of population is introduced to imitate the detection behavior of fruit fly. The experiment on 12 benchmark functions shows that MGFOA is more effective and robust than basic FOA, Global Particle Swarm Optimization Algorithm (GPSO) and another improved FOA (LGMS-FOA).