Hybrid Fuzzy-Particle Swarm Optimization-Simplex (F-PSO-S) Algorithm for Optimum Design of PM Drive Couplings

A. El-Wakeel, A.C. Smith

    Research output: Contribution to journalArticlepeer-review

    Abstract

    This paper aimed at exploring the use of the proposed hybrid Fuzzy-Particle Swarm Optimization-Simplex (F-PSO-S) algorithm to optimize the structural design of PM couplings subject to several key design constraints. The new proposed hybrid optimization algorithm is constructed based on combining three well-known techniques: fuzzy logic (FL), particle swarm optimization (PSO), and simplex method (SM). FL is used to aggregate different scaling and/or conflicting objectives in one objective function using fuzzy combination operators. On the other hand, the PSO has obvious capabilities in global search whereas the SM has exceptional advantages in local search. As a hybrid algorithm, the F-PSO-S has the outstanding feature of combining the ability of global searching and local canvassing for different scales and/or conflicting objective functions. The proposed algorithm has been applied to permanent magnet (PM) drive couplings. A standard coupling design is used as a good initial point for the conventional SM and to define the performance constraints for the proposed optimization technique. New coupling designs are developed and optimized to show the superior capabilities of the F-PSO-S algorithm as a global optimization technique. The sensitivity analysis is also performed to identify the effects of different design parameters on the coupling performance.
    Original languageEnglish
    Pages (from-to)1560-1571
    Number of pages11
    JournalElectric Power Components and Systems
    Volume43
    Issue number13
    DOIs
    Publication statusPublished - Jul 2015

    Keywords

    • Permanent magnet couplings

    Fingerprint

    Dive into the research topics of 'Hybrid Fuzzy-Particle Swarm Optimization-Simplex (F-PSO-S) Algorithm for Optimum Design of PM Drive Couplings'. Together they form a unique fingerprint.

    Cite this