【摘要】 针对烟花算法收敛速度慢和求解精度不高,论文提出了一种改进烟花算法——带柯西变异的自学习改进烟花算法.改进算法用全局搜索能力更强的柯西变异算子替代高斯变异算子,增大变异范围;用全局最优烟花个体和历史柯西火花的位置来构造新的爆炸半径使其不仅能够继承和学习历史信息,还能够自适应地调整步长;并使用可同时兼顾烟花质量与分布的"精英-随机"选择策略.使用了10个典型基准测试函数和10个0-1背包问题进行仿真实验,结果表明,与蝙蝠算法、粒子群算法、带高斯扰动的粒子群算法、烟花算法、增强烟花算法、自适应烟花算法相比.该算法在收敛速度、计算精度以及稳定性方面性能更优.更多还原

【Abstract】 Aiming at the problems of slow convergence speed and low precision of Fireworks Algorithm,this paper proposes an improved algorithm of fireworks algorithm,which uses cauchy mutation with stronger global search ability to replace gaussian mutation,so as to increase the range of variation and improve the global search ability. The new explosion radius is constructed by using the location information of globally optimal fireworks individuals and historical cauchy sparks,so that it can not only inherit and learn historical information,but also adjust the step size adaptively. An"elite-random"selection strategy is proposed,which can give consideration to both the quality and distribution of fireworks. Simulation results of ten typical benchmark functions and ten 0-1 knapsack problems show that,compared with Bat Algorithm( BA),Particle Swarm Algorithm( SPSO),Particle Swarm Algorithm with Gaussian Mutation(GPSO),Fireworks Algorithm( FWA),Enhanced Fireworks Algorithm( EFWA),and Adaptive Fireworks Algorithm( AFWA).The algorithm has better convergence speed,accuracy and stability.更多还原