| Summary: | This work introduces new enhancements to the Bees Algorithm in order to improve its overall performance. These enhancements are early neighbourhood search process, efficiency based recruitment for neighbourhood search process, hybrid strategy involving tabu search, new escape mechanism to escape locals with similar fitness values and autonomy to minimise interaction between search process and the user. The proposed enhancements were applied alone or in pair to develop improved versions of the Bees Algorithm. Three Enhanced Bees Algorithms were introduced: the Early Neighbourhood Search and Efficiency Based recruitment Bees Algorithm (ENSEBRBA), the Hybrid Tabu Bees Algorithm (TBA) and the Autonomous Bees Algorithm (ABA). The ENSEBRBA with an empowered initialisation stage and extra recruitment for neighbourhood search is introduced to improve performance of the Bees Algorithms on high dimensional problems. The TBA is proposed as a new version of the Bees Algorithm which utilises the memory lists to memorise less productive patches. Moreover, the local escape strategy was also implemented to this algorithm. Proposed modifications increased the productivity of the Bees Algorithm by decreasing number of evaluations needed to converge to the global optimum. iii The ABA is developed to provide independency to the Bees Algorithm, thus it is able to self tune its control parameters in a sub-optimal manner. All enhanced Algorithms were tested on continuous type benchmark functions and additionally, statistical analysis was carried out. Observed experimental results proved that proposed enhancements improved the Bees Algorithm’s performance.
|
|---|
