一种基于反馈和共享机制的自适应进化策略

Adaptive evolutionary strategy based on feedback and sharing mechanism

为提高进化策略的搜索精度和全局搜索能力,提出了一种基于反馈和共享机制的改进进化策略,将各代当前最优搜索结果和反映种群个体聚集程度的共享度信息反馈到变异步长的更新式中,以保证种群中的部分个体在搜索后期仍保持较大的跳出局部极小的能力,从而达到提高算法全局搜索能力和搜索精度的目的。为了对比改进后进化策略与常规进化策略的优化效果,利用三个测试函数对两种进化策略进行了仿真测试。测试结果表明,与常规进化策略相比,提出的基于反馈和共享机制的进化策略具有更高的搜索精度和更强的全局搜索能力。

In order to improve the searching precision and the global searching ability of evolutionary strategy （ES）, this paper proposed an adaptive evolutionary strategy based on the feedback and sharing mechanism （ESBFSM）, in which the information of every generation＇s current optimal searching result and sharing degree was fed into the mutating formula. By this way, a part of individuals of the population in the later searching procedure kept higher probability of jumping out the local minimum. To compare the optimizing effect between traditional ES and the ESBFSM, the test based on three benchmarks were done. The test result demonstrates that the ESBFSM has higher global searching ability and precision than the traditional ES.