基于信息熵的混合引力搜索算法

针对基本引力搜索算法搜索速度慢和容易出现早熟的缺点,本文提出了一种基于信息熵的混合引力搜索算法. 受粒子群算法的启发,所提算法首先通过改进基本引力搜索算法的速度和位置更新公式来提高搜索速度；其次,通过惯性质量构造了信息熵模型来刻画种群的寻优程度,并采用不同的信息熵阈值动态选择权重,平衡了算法的全局搜索能力和局部搜索能力. 用8个标准测试函数的仿真实验和基本引力搜索算法与记忆改进的引力搜索算法的比较表明了所提算法收敛速度快,鲁棒性强且效率高.

A hybrid gravitational search algorithm with information enproty

Based on Information entropy, this paper proposes a hybrid gravitational search algorithm to overcome drawbacks of basic gravitational search algorithm, such as low speed, prematurity, and so on. Inspired by particle swarm optimization, the proprosed algorithm first modifies the formulate of velocity and position to enhance the searching speed. To balance the ability of global search and local search, by means of mass, information entropy model is then built to characterize the population optimization degree, and different weights are chosen according to different threshold of the information entropy. Numerical experiments on eight benchmark functions illustrate that the proposed is superior to basic gravitational search algorithm(GSA) and memory gravitational search algorithm(MGSA).