Artificial super-Bee enhanced Colony (AsBeC) algorithm for numerical optimization with limited function evaluations Part 1: technologies and benchmark validation

The Artificial Bee Colony (ABC) algorithm is a very effective, simple and robust nature-based metaheuristic optimization procedure. The standard ABC has been recently developed and it currently raises a lot of interest and upgrading efforts, since it finds attractive applications in many fields of the scientific research. One of these application is represented by optimizations that employ very expensive numerical simulations in terms of resources. In this framework there is a strong need for methods that assure the best improvement with the shortest analyses time. In this paper, the authors propose and recall several modifications to the standard ABC in order to improve its speed and solution accuracy for problems where function evaluations have to be limited around 103. These technologies consist in enhancements of the basic structure and hybridizations with other optimization strategies. Moreover three different kinds of parallelization in the code are analysed. Each modification as well as their combinations are studied and explained; the performance of modified ABC is evaluated through extensive simulations over different analytical test functions. Moreover, standard settings for this new algorithm, called Artificial super-Bee enhanced Colony (AsBeC), are given in order to maintain the simplicity of ABC. The present article explores all the technical issues involved about AsBeC, while the second part will consider a real-like application to the engineering optimal design of turbomachinery through Computational Fluid Dynamics, environment in which the improved algorithm was originally conceived by the authors.