一种改进鱼鹰优化算法及其应用
An improved osprey optimization algorithmand its application
陈曦明 1张军伟 1张冉 1杨波 1吴学雷 1刘浩 1毕一白1
作者信息
- 1. 中国运载火箭技术研究院 北京航天发射技术研究所,北京 100076
- 折叠
摘要
针对鱼鹰优化算法(osprey optimization algorithm,OOA)在运行时存在寻优精度和稳定性差的问题,提出了如下改进策略:首先,将SPM混沌映射融入种群的初始化阶段,提升种群多样性;其次,在勘探和开采阶段分别利用威布尔分布的长、短距离随机扰动策略更新鱼鹰的位置,可有效改善OOA的收敛精度;最后,提出一种"最优-随机均值"的变异策略,用于强化OOA迭代过程中跳出局部最优的能力.所提出的算法称为改进鱼鹰优化算法(improved osprey opti-mization algorithm,IOOA).为了验证IOOA的寻优能力,将IOOA与其他新兴智能算法分别对12个基准函数进行寻优对比实验,结果表明:IOOA的寻优成功率、收敛速度以及稳定性显著高于其他算法.此外,将所提出的IOOA应用于混合核相关向量机的超参数寻优中,进一步用于柴油机多目标性能预测.
Abstract
To address the poor accuracy and stability for Osprey Optimization Algorithm ( OOA) , this paper proposes some improvement strategies. First, SPM chaotic mapping has been integrated into the stage of population initialization to improve the population diversity. Second, Weibull's long and short distance random disturbances are integrated respectively in the exploration and mining stages to update the position of osprey, effectively improving the convergence accuracy of OOA. Finally, a mutation strategy of"optimum-random mean"is proposed to enhance the ability of algorithm to jump out of the local optimal during the iterative process. The proposed algorithm is called Improved Osprey Optimization Algorithm ( IOOA) . To verify the optimization ability of IOOA, it is compared with other emerging intelligent algorithms for the optimization of 12 benchmark functions. Our results show the success rate of optimization, convergence speed and stability of IOOA are significantly higher than those of other algorithms. In addition, the application of IOOA on the hyperparameter optimization of Hybrid Kernel Relevance Vector Machine is able to accurately predict the multi-objective performance of diesel engines.
关键词
鱼鹰优化算法/SPM混沌映射/威布尔随机扰动/最优-随机均值变异策略Key words
osprey optimization algorithm/SPM chaotic mapping/weibull random disturbance/optimum-random mean mutation引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据