A novel dual-biological-community swarm intelligence algorithm with a commensal evolution strategy for multimodal problems

• Published in: The Journal of supercomputing Vol. 77; no. 10; pp. 10850 - 10895
• Main Authors: Ren, Hui, Shen, Xiaochen, Jia, Xiaojun
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.10.2021; Springer Nature B.V
• Subjects: Algorithms; Compilers; Computer Science; Convergence; Data integration; Deep Learning in IoT: Emerging Trends and Applications - 2019; Evolutionary algorithms; Internet of Things; Interpreters; Multisensor fusion; Particle swarm optimization; Performance evaluation; Processor Architectures; Programming Languages; Swarm intelligence; Wireless sensor networks; Swarm intelligence algorithm; Internet of Things (IoT); Commensal evolution strategy; Multimodal problems
• ISSN: 0920-8542; 1573-0484
• DOI: 10.1007/s11227-021-03721-8

For sophisticated applications, engineers should always consider multi-objective, multi-task or multi-modal problems, especially in the Internet of Things, such as the data fusion of multi-sensor systems, multiple routing problems for automation and convergence control in wireless sensor networks. Normally, utilizing traditional methods costs abundant resources and yields the homogeneous results that fails to satisfy requirements. For the multimodal situation, this paper proposes the dual-biological-community swarm intelligence algorithm based on particle swarm optimization (DBC–PSO), which is combined with the commensal evolution strategy to enhance the convergence ability. This algorithm can split tasks into two communities and guarantee regional changes in information and search accuracy for multimodal problems through a commensal strategy. Moreover, some typical parameters, including the population rate, velocity and radius influence, are considered to study the algorithm performance. The performance is evaluated on 12 well-known multimodal problems, and the simulation is compared with some algorithms that utilize a similar evolutionary approach. The results indicate that the proposed algorithm exhibits strong performance and is very promising for use in more productive work.