A modified ABC algorithm based on improved-global-best-guided approach and adaptive-limit strategy for global optimization

• A modified ABC algorithm is proposed for global optimization problems.
• Improved global-best-guided term with nonlinear adjusting factors is employed.
• Nonlinear adjusting factors are used to balance the exploration and exploitation.
• An adaptive-limit strategy is designed to control the variable limit.
• Multiple numerical experiments are conducted to verify the proposed algorithm.

Artificial bee colony (ABC) is a novel biological-inspired optimization algorithm which has been shown to be more effective for global optimization of multimodal and multidimensional optimization problems, than some other conventional biological-inspired optimization algorithms, such as genetic algorithm (GA) and particle swarm optimization (PSO), for its good exploration capability and the efficient balance between the local search and the global search processes. It has drawn widely attentions from scholars and was applied to various fields for its advantages of excellent global optimization ability and it is easy to implement. However, the basic ABC has some drawbacks such as poor exploitation, slow to converge and hard to find the best solution from all feasible solutions in some cases. In this paper, a modified ABC algorithm based on improved-global-best-guided approach and adaptive-limit strategy for global optimization problems called IGAL-ABC algorithm is proposed. An improved-global-best-guided term with a nonlinear adjusting factor is employed in the update equation. Two nonlinear adjusting factors are applied to control the convergence speed and balance the exploration and exploitation abilities. Multiple dimensions of solution are perturbed each time for generating new candidate food sources. In addition, an adaptive-limit strategy is applied to adjust the limit which controls the frequency that the employed bee abandons its food source, to improve the performance of the algorithm further. Results of experiments tested on multiple benchmark functions show that the proposed method is effective and has good performance. The comparison experiments illustrate that the proposed algorithm has better solution quality and convergence characteristics.

Figure optionsDownload as PowerPoint slide