Median-oriented bat algorithm for function optimization

Abstract

The bat algorithm is easily trapped into local optima, the population diversity is poor, and the optimizing precision is bad. In order to overcome these disadvantages, this paper presents a median bat algorithm (MBA) to avoid local optima and carry out a global search over entire search space. The proposed algorithm adopts the median position of the bats. And the median and worst bats are combined to the basic bat algorithm to achieve a better balance between the global search ability and local search ability. The simulation results of 10 standard benchmark functions show that the proposed algorithm is effective and feasible in both low-dimensional and high-dimensional case. Compared to the basic bat algorithm, particle swarm optimization and CLSPSO, the proposed algorithm can get high precision and can almost reach the theoretical value.