计算机科学与探索 ›› 2020, Vol. 14 ›› Issue (10): 1787-1800.DOI: 10.3778/j.issn.1673-9418.1909060

• 理论与算法 • 上一篇    

异质空间结构种群迁徙动力学优化算法

黄光球,陆秋琴   

  1. 西安建筑科技大学 管理学院,西安 710055
  • 出版日期:2020-10-01 发布日期:2020-10-12

Heterogeneous Spatial Structure-Based Population Migration Dynamics Optimization Algorithm

HUANG Guangqiu, LU Qiuqin   

  1. School of Management, Xi'an University of Architecture and Technology, Xi'an 710055, China
  • Online:2020-10-01 Published:2020-10-12

摘要:

为了求解一些复杂优化问题,采用异质空间结构种群迁徙动力学理论,提出了异质空间结构种群迁徙动力学优化算法(HSS-PMDO)。在该算法中,优化问题的解空间与由若干个斑块组成的海岛相对应,每个斑块具有竞争、互利和捕食-被食3个生存条件之一。种群会依据所在斑块的生存条件选择适应度更好的斑块进行迁徙。在斑块上,种群相互之间展开与生存条件相适应的相互作用。依据种群的迁徙和相互作用开发出了5个算子:竞争算子可提升算法的求精能力;互利算子和捕食-被食算子可提升算法的探索能力;迁徙算子可使得种群间信息交换充分,从而提升了探索能力和求精能力的平衡性;选择算子可确保算法具有全局收敛性。当种群不断进化时,每次只有其部分特征发生变化,该特征可提高收敛速度。测试结果表明,HSS-PMDO算法具有收敛速度快,探索、求精及其平衡能力强,能够快速求解一些维数较高的复杂优化问题。

关键词: 群智能优化算法, 全局优化, 异质空间结构种群迁徙动力学理论

Abstract:

To solve some complex optimization problems, an optimization algorithm of population migration dynamics with heterogeneous spatial structure (HSS-PMDO) is proposed by using the theory of population migration dyna-mics with heterogeneous spatial structure. In this algorithm, the solution space of the optimization problem corre-sponds to the island composed of several patches, each of which has one of the three survival conditions of compe-tition, mutual benefit and predator-prey. Populations will choose the patch with better adaptability for migration according to the living conditions of the patch. On the patch, populations interact with each other to adapt to the living conditions. Five operators are developed according to population migration and interaction. The competition operator can improve the exploitation ability of the algorithm. The mutual benefit operator and predator-prey ope-rator can improve the exploration ability of the algorithm. Migration operator can make the information exchange among populations sufficient, so as to improve the balance of exploration ability and exploitation ability. The selec-tion operator can ensure the global convergence of the algorithm. When the population evolves continuously, only part of its features change each time. The characteristic can improve the convergence speed. The test results show that the HSS-PMDO algorithm has fast convergence speed, strong exploration, exploitation and balance ability, and can quickly solve some complex optimization problems with higher dimensions.

Key words: swarm intelligence optimization algorithm, global optimization, population migration dynamics theory with heterogeneous spatial structure