An improved monkey algorithm with dynamic adaptation

• Chaotic search method is utilized to generate the random numbers in MA.
• Step length is adjusted dynamically according to evolutionary speed factor.
• Eyesight and watch times grow adaptively as the aggregation degree increases.
• Somersault distance increases with the increase of the aggregation degree.
• Modified algorithm improves search ability of MA and has an advantage of robustness.

The paper proposes an improved monkey optimization algorithm with dynamic adaptation. In the algorithm, the chaotic search method instead of random process is employed to generate random numbers in initialization process, climb process and watch-jump process so as to avoid repeating search in the same neighborhood. Moreover, two parameters (the evolutionary speed factor and the aggregation degree) are introduced to describe the evolution state of monkeys. The step length of each monkey in climb process is different and decreases dynamically based on its own evolutionary speed factor, and an increase of the aggregation degree of monkeys achieves the dynamical increase of the eyesight, watch times and somersault distance of monkeys. Then, features of the improved algorithm are analyzed through performing several test functions in numerical experiments. Finally, experimental results show that the algorithm significantly improves the search ability of monkey algorithm and has an advantage of robustness.