Enhancing the bees algorithm using the traplining metaphor

Ismail, Asrul Harun ORCID: 0000-0002-6580-4654 (2021). Enhancing the bees algorithm using the traplining metaphor. University of Birmingham. Ph.D.

Preview

Ismail2021PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (7MB) | Preview

Abstract

This work aims to improve the performance of the Bees Algorithm (BA), particularly in terms of simplicity, accuracy, and convergence. Three improvements were made in this study as a result of bees’ traplining behaviour.

The first improvement was the parameter reduction of the Bees Algorithm. This strategy recruits and assigns worker bees to exploit and explore all patches. Both searching processes are assigned using the Triangular Distribution Random Number Generator. The most promising patches have more workers and are subject to more exploitation than the less productive patches. This technique reduced the original parameters into two parameters. The results show that the Bi-BA is just as efficient as the basic BA, although it has fewer parameters.

Following that, another improvement was proposed to increase the diversification performance of the Combinatorial Bees Algorithm (CBA). The technique employs a novel constructive heuristic that considers the distance and the turning angle of the bees’ flight. When foraging for honey, bees generally avoid making a sharp turn. By including this turning angle as the second consideration, it can control CBA’s initial solution diversity.

Third, the CBA is strengthened to enable an intensification strategy that avoids falling into a local optima trap. The approach is based on the behaviour of bees when confronted with threats. They will keep away from re-visiting those flowers during the next bout for reasons like
predators, rivals, or honey run out. The approach will remove temporarily threatened flowers from the whole tour, eliminating the sharp turn, and reintroduces them again to the habitual tour’s nearest edge. The technique could effectively achieve an equilibrium between exploration and exploitation mechanisms. The results show that the strategy is very competitive compared to other population-based nature-inspired algorithms.

Finally, the enhanced Bees Algorithms are demonstrated on two real-world engineering problems, namely, Printed Circuit Board insertion sequencing and vehicles routing problem.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):