A novel gravitational search algorithm with modified step-size and Explosion-Birth operator
Donya Yazdani, Mohammad Reza Meybodi · 2014
Gravitational Search Algorithm (GSA) is a population-based optimization algorithm based on Newton's law of gravity and the notion of mass interactions. GSA has the advantage of proper global search ability. However, it suffers from weak local search due to relatively big step-size of agents in the search process. In order to improve the balance between exploration and exploitation of GSA, two mechanisms are introduced in this paper. First, a modified velocity formula (for step-size computation) is proposed which improves the exploitation ability of the GSA significantly and is able to set a proper balance between exploitation and exploration. Second, a new nature-inspired operator named Explosion-Birth operator is introduced to add alongside the first mechanism to GSA for speeding up the convergence and improving exploitation ability. This operator is based on explosion and birth of stars in the galaxy. The experiments are conducted on 10 standard benchmarks including unimodal and multimodal functions. The experimental results confirm the high performances of both proposed mechanisms compared with those of standard GSA and four other well-known algorithms in this field.