一种自适应调整种群子代数量 与步长的优化算法——爆米花算法

提出了一种新的群体智能优化算法——爆米花算法。借鉴了烟花算法爆炸机制的优点,利用个体在寻优过程中适应度值的优劣来动态调整子代的数量,个体的适应度值越好,产生的子代数量越多,并且在该个体附近搜索的子代数量越多,以此控制局部搜索与全局搜索之间的平衡。还借鉴了粒子群优化算法的记忆机制,引入个体最优和全局最优来构造新的爆炸半径,使算法能够在寻优过程中动态地调整步长,并对全局最优进行高斯扰动,增加种群的多样性。实验结果表明:与其他优化算法(如蝙蝠算法、标准粒子群算法、烟花算法)相比,本文提出的爆米花算法总体性能更优。

Popcorn algorithm: A self-adaptive algorithm for adjusting population of offspring and optimizing step size

We present a new swarm intelligent optimization algorithm called popcorn algorithm. The popcorn algorithm learns from the advantage of the explosion mechanism of the fireworks algorithm and takes advantage of the individual particle’s fitness value in the optimization process to adjust the number of offspring dynamically. The better the individual particle’s fitness value is, the larger of the offspring population, and the more of the offspring searching in the vicinity of the individual particle. The algorithm adjusts the number of offspring dynamically to control the balance between local search and global search. In addition, it uses the memory mechanism of the particle swarm optimization algorithm as reference, and introduces the best individual particle and the best global particle to construct a new explosion radius, so that it can adjust the step size dynamically in the optimization process. The Gaussian perturbation is performed on the best global particle to increase the diversity of the population. Experimental results show that compared with other optimization algorithms such as the bat algorithm, standard particle swarm optimization algorithm and fireworks algorithm, the overall performance of the proposed popcorn algorithm is better.