改进藤壶配尾优化算法求解高维连续优化问题

PDF下载HTML

Title:

Improved barnacles mating optimizer to solve high-dimensional continuous optimization problems

作者:

赵世杰^1,2, 张天然¹, 马世林¹, 王梦晨¹

1. 辽宁工程技术大学 智能科学与优化研究所, 辽宁 阜新123000;
2. 辽宁工程技术大学 运筹与优化研究院, 辽宁 阜新123000

Author(s):

ZHAO Shijie^1,2, ZHANG Tianran¹, MA Shilin¹, WANG Mengchen¹

1. Institute of Intelligence Science and Optimization, Liaoning Technical University, Fuxin 123000, China;
2. Institute for Optimization and Decision Analytics, Liaoning Technical University, Fuxin 123000, China

关键词:

智能优化算法; 藤壶优化算法; 沉降附着行为; 正反向递减铸型策略; 局部极值规避; 高维函数优化; 全局寻优; 收敛精度

Keywords:

intelligence optimization algorithm; barnacles mating optimizer; sedimentation adhesion behavior; forward-and-backward decreasing casting strategy; local extremum avoidance; high dimensional optimization of function; global optimization; convergence precision

分类号:

TP391; TP301.6

DOI:

10.11992/tis.202110022

摘要:

为增强藤壶配尾优化算法(barnacles mating optimizer, BMO)的全局探索性能和局部寻优精度，融合藤壶幼虫的沉降附着行为与正反向递减铸型策略提出一种改进藤壶配尾优化算法(improved BMO, IBMO)并将其用于求解高维连续优化问题。沉降附着行为模型受自然界藤壶幼虫随潮浮游、螺旋沉降的行为启发所构建，以增加种群多样性并改善算法的全局探索性能。正反向递减铸型策略借鉴反向学习思想并融入递减调控机制修正传统藤壶优化算法的精子铸型过程，以扩增种群的局部搜索域并改善算法的局部开采性能。实验结果表明，两种策略可分别有效改善藤壶优化算法的全局探索和局部开采性能；同时，所提IBMO算法相较于其他新近智能算法则表现出更高收敛精度、更强算法稳健性和良好高维适用性等。

Abstract:

To strengthen the global exploration performance and local optimization accuracy of barnacles mating optimizer (BMO), an improved BMO (IBMO) is proposed based on the sedimentation adhesion behavior (SAB) of barnacle larva and the forward-and-backward decreasing casting (FBDC) strategy, which is applied to solve high-dimensional continuous optimization problems. Inspired by the behavior of barnacle larva floating with tide and spiraling sedimentation in nature, the SAB model is built to increase the population diversity and improve the global exploration capacity. Meanwhile, in accordance with reverse learning, and by integrating into the decreasing control mechanism, FBDC modifies the sperm casting process of traditional BMO to amplify the local search domain and improve the local exploitation ability. Experimental results verify that these two strategies can effectively improve the global exploration and local optimization exploitation performance of BMO. Compared with other recent intelligence algorithms, the proposed IBMO shows higher convergence accuracy, stronger robustness and good high-dimensionality applicability.