A multi-granularity genetic algorithm

The genetic algorithm is a classical evolutionary algorithm that mainly consists of mutation and crossover operations. Existing genetic algorithms implement these two operations on the current population and rarely use the spatial information that has been traversed. To address this problem, this paper proposes an improved genetic algorithm that divides the feasible region into multiple granularities. It is called the multi-granularity genetic algorithm (MGGA). This algorithm adopts a multi-granularity space strategy based on a random tree, which accelerates the searching speed of the algorithm in the multi-granular space. Firstly, a hierarchical strategy is applied to the current population to accelerate the generation of good individuals. Then, the multi-granularity space strategy is used to increase the search intensity of the sparse space and the subspace, where the current optimal solution is located. The experimental results on six classical functions demonstrate that the proposed MGGA can improve the convergence speed and solution accuracy and reduce the number of calculations required for the fitness value.