求解高维函数优化的混合智能算法

摘要

A kind of hybrid intelligent algorithm is proposed according to the study on the several high dimensional functions optimization. Applying the evolution system of the population migration algorithm with simplified steps, based on cloud model’s cloud droplets of randomness and stable tendentiousness, the initial population is replaced by cloud model’s cloud droplets, and population flow is converted into the process that a new generation of cloud droplets is produced by the previous ones. Using the strong disturbance of Cauchy distribution, Cauchy variation is applied to the population in the preferential area to prevent the search trapping into local extreme value. The simulation results show that the algorithm has high solving quality and stable performance, for several typical super high dimensional functions, which the dimension is as high as 10000, the algorithm can still find the theoretical optimal value.%针对几个高维函数优化进行了研究，提出一种混合智能算法。借鉴人口迁移算法的进化体制，精简了算法步骤；鉴于云模型的云滴在随机中带有稳定倾向性，将人口进化过程中初始人口群体由云模型的云滴代替，人口流动转化为上一代云滴产生新一代云滴的过程；为防止寻优陷入局部极值，借用柯西分布的强扰动性，对优惠区域的人口实施柯西变异。几个典型高维函数的仿真实验表明，算法求解质量高、性能稳定，甚至对几个维数高达10000维的超高维函数，算法都可以稳定收敛到理论最优。