An improved bat algorithm hybridized with extremal optimization and Boltzmann selection

As a meta-heuristic algorithm, bat algorithm (BA) is based on the characteristics of bat-based echolocation and has been widely used in various aspects of optimization problems since it appeared. However, the original BA still has many shortcomings, such as insufficient local search ability, lack of diversity and poor performance on high-dimensional optimization problems. To overcome these weaknesses, this paper proposes an improved BA with extremal optimization (EO) algorithm (IBA-EO) to improve the performance of BA. In IBA-EO, an improved update strategy is proposed to obtain the solutions generating from the random selected bats to enhance the global search capability. The exploitation ability is improved by EO algorithm with excellent local search capability. Furthermore, Boltzmann selection and a monitor mechanism are employed to keep suitable balance between exploration ability and exploitation ability. To testify the performance of IBA-EO in handling various optimization problems, this study considers four groups of contrast experiments. Extensive simulation results demonstrate that IBA-EO can achieve a strong competitive performance by comparing with other fifteen well-established algorithms in terms of accuracy, reliability and statistical tests.