A chaotic ergodicity based evolutionary computation algorithm

We propose a novel population-based optimization algorithm, Chaotic Evolution (CE), that uses a chaotic ergodicity to implement exploitation and exploration functions of the evolutionary computation algorithm. A new control parameter, direction factor rate, is proposed in CE to guide search direction. Compared with differential evolution (DE), our proposal works with the more simple principle, and can obtain the better optimization performance, escape from the local optimum and avoid the premature. By changing the chaotic system in our proposal, it is easy to extend its search capability, i.e., the scalability of our proposal is higher than DE. A series of comparative evaluations are conducted to analyze the feature of the proposal. From these results and analysis, our proposed algorithm can optimize most of benchmark functions and outperforms better than DE.