解决高维优化问题的差分进化算法

Differential evolution algorithm for high dimensional optimization problem

针对高维优化问题难以解决并且优化耗费时间长的问题,提出了一种解决高维优化问题的差分进化算法。将协同进化思想引入到差分进化领域,采用一种由状态观测器和随机分组策略组成的协同进化方案。其中,状态观测器根据搜索状态反馈信息适时地调用随机分组策略重新分组;随机分组策略将高维优化问题分解为若干较低维的子问题,而后分别进化。该方案有效地增强了算法解决高维优化问题的搜索速度和搜索能力。经典型的实例测试,并与其他一流差分进化算法比较,实验结果表明:所提算法能有效地求解不同类型的高维优化问题,在搜索速度方面有明显提升,尤其对可分解的高维优化问题极具竞争力。

In order to solve the problem that high dimensional optimization problem is hard to optimize and time- consuming, a Differential Evolution for High Dimensional optimization problem （DEHD） was proposed. By introducing coevolutionary to differential evolution, a new coevolution scheme was adopted, which consisted of state observer and random grouping strategy. Specifically, state observer activated random grouping strategy according to the feedback of search status while random grouping strategy decomposed high dimensional problem into several smaller ones and then evolved them separately. The scheme enhanced the algorithm＇s search speed and effectiveness. The experimental results show that the proposed algorithm is effective and efficient while solving various high dimensional optimization problems. In particular, its search speed improves significantly. Therefore, the proposed algorithm is competitive on separable high dimensional problems.