Pusat Pengajian Kejuruteraaan Elektrik dan Elektronik - Tesis

Browse

Browsing Pusat Pengajian Kejuruteraaan Elektrik dan Elektronik - Tesis by Author "Abdul Karim, Aasam"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Publication
    Modified particle swarm optimization for single objective problems in continuous space
    (2022-04-01)
    Abdul Karim, Aasam
    Particle Swarm Optimization (PSO) is a prominent Swarm Intelligence (SI) algorithm which achieves significant optimization performance for global optimization problems. However, slow convergence, local optima entrapment and improper balance in exploration and exploitation searches may occur if it is applied to solve complex problems. To overcome these limitations, this research has proposed three improved PSO variants, namely, Vital Information Selector PSO (VISPSO), Effective Guide PSO (EGPSO) and Hovering Swarm PSO (HSPSO). To strengthen the exploration and exploitative searches of the swarm, the VISPSO guides its swarm using an optimized guide particle created using the directional information contained in the two nearest neighbors of the global best particle. The EGPSO is proposed to improve the VISPSO by intelligently providing an alternate search trajectory to those particles stuck in local optima. The HSPSO further improves the VISPSO and EGPSO by using a two swarm approach and creating a cooperative mechanism which realizes an effective division of labor among the particles of the two subswarms. The optimization performances of these three proposed PSO variants have been compared with six conventional PSO variants using 2014 IEEE Congress on Evolutionary Computation (CEC 2014) and four real-world engineering design problems. The experimental results obtained by each proposed PSO variant are also thoroughly evaluated and verified via the non-parametric statistical analyses. Based on the experiment results, the VISPSO exhibits better search accuracy, search reliability, and search efficiency in solving different benchmark functions. The EGPSO shows significant performance improvement over the VISPSO, however, the computational cost of the algorithm is compromised. On the other hand, the robustness of the HSPSO is not effected by the challenging optimization problems with different characteristics. The study concludes that the optimization performance of HSPSO is much better than VISPSO, EGPSO and other conventional PSO variants as it is able to improve the exploration and exploitation search capabilities with lower processing time.