反向烟花算法及其应用研究
Backward Fireworks Algorithm and Application Research

针对烟花算法性能提升瓶颈和收敛速度较慢的问题,通过引入反向学习策略,提出了一种自适应反向学习算子,并进行了相关收敛性理论证明。通过反向学习算子与烟花算法相结合,构建了反向烟花算法组,并通过典型测试函数进行仿真实验。结果表明:在相同实验设置下,反向烟花算法可在原算法寻优性能上至少提升10－2精度,并加快了收敛速度。针对混沌同步与控制系统中常见的参数辨识问题,以混沌同步控制中Lorenz混沌系统参数辨识问题为应用背景,通过实验仿真,验证了反向烟花算法可用于混沌控制系统参数估计,与现有方法相比较,估计误差低至10－11,具有较高的估计精度,是一种新的有效的混沌控制系统参数估计方法,拓展了算法工程应用的范围。
Aiming at the performance bottlenecks and slow convergence of fireworks algorithm (FWA), an adaptive backward learning operator (ABLO) is proposed through the introduction of backward learning strategy, and its convergence performance is proved theoretically. By combining FWA with ABLO, a set of hybrid FWA is proposed and verified by typical test functions. The results show that under the same experimental setup, the backward fireworks algorithm can improve the computation accuracy by at least 10－2 in the optimization performance of original algorithm and the convergence rate is enhanced. Finally the algorithm is applied to identify the parameters of Lorenz chaotic system. Through simulation experiments, it is verified that this algorithm can be used for parameter identification of chaotic control systems. Compared with other swarm intelligence algorithms, its identification error is as low as 10－11. It is a novel and effective parameter identification method for chaotic control systems