«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2025年01期

页码:

140-147

栏目:

人工智能

出版日期:

2025-01-10

文章信息/Info

Title:

Improved Red-tailed Hawk Optimizer Integrating Multiple Strategies and Its Applications

文章编号:

1673-629X(2025)01-0140-08

作者:

赵倩; 郭锋*; 王婷; 刘元凯

临沂大学 信息科学与工程学院,山东 临沂 276000

Author(s):

ZHAO Qian; GUO Feng*; WANG Ting; LIU Yuan-kai

School of Information Science and Engineering,Linyi University,Linyi 276000,China

关键词:

红尾鹰优化算法; 佳点集; 切线飞行; 高斯柯西变异; 光伏功率预测; 极限学习机

Keywords:

red - tailed hawk optimizer;  best point set;  tangential flight;  Cauchy Gaussian variation;  PV power prediction;  extreme learning machine

分类号:

TP18

DOI:

10.20165/j.cnki.ISSN1673-629X.2024.0276

摘要:

红尾鹰优化算法(Red-tailed Hawk Optimizer,RTH)具有较强的全局搜索和快速收敛能力,但收敛精度低、局部开发性能不足。 针对此问题,该文提出一种融合多策略改进的红尾鹰优化算法( Improved Red - tailed Hawk Optimizer,IRTH)。 首先,在初始种群生成阶段引入了佳点集模型,以增进红尾鹰种群的多样性;其次,利用切线飞行策略优化步长来改进个体的位置更新策略,从而提高算法的收敛速度和精度;最后,引入高斯柯西变异策略,解决算法迭代后期容易陷入局部最优问题。 采取基准测试函数、Wilcoxon 秩和检验等实验将改进算法 IRTH 与其他 5 种经典算法进行对比,实验结果表明,IRTH 收敛性和寻优精度整体均优于其他算法。 最后,将 IRTH 应用于光伏功率预测领域,结果表明,该模型的RMSE、MAPE 和 MAE 分别为 9. 58% 、0. 61% 和 6. 91% ,相比单一极限学习机模型和其他算法优化模型分别提高了 85.04% 、22.48% 和 67% ,从实际工程领域验证了改进策略的有效性和优异性。

Abstract:

Red-tailed Hawk Optimizer (RTH) has strong global search and fast convergence ability,but low convergence accuracy and insufficient local development performance. To solve this problem,we propose an Improved Red-tailed Hawk Optimizer ( IRTH) that integrates multiple strategies for enhancement. Firstly,the best point set model is introduced in the initial population generation stage to enhance the diversity of the red-tailed hawk population. Secondly,tangential flight strategy is used to optimize the step size and improve the individual position update strategy, so as to improve the convergence speed and accuracy of the algorithm. Finally, the Cauchy Gaussian variation strategy is introduced to solve the problem that the algorithm is prone to fall into local optimal in the late iteration. The benchmark test function and Wilcoxon rank sum test are used to compare the improved IRTH algorithm with other five classical algorithms. The experimental results show that the convergence and optimization accuracy of IRTH are better than that of other algorithms. Finally,IRTH is applied to the PV power prediction field. It is showed that RMSE,MAPE and MAE of the model are 9.58% ,0. 61% and 6. 91% , respectively, which are 85. 04% ,22. 48% and 67% higher than that of the single Extreme Learning Machine and other algorithm optimization models,respectively. The effectiveness and excellence of the improved strategy are verified from the practical engineering field.

常用功能

更新日期/Last Update: 2025-01-10