Artificial Bee Colony Algorithm Combined With Previous Successful Search Experience

The neighboring solutions-based technique employed by the artificial bee colony algorithm (ABC) is good at exploration but poor at exploitation. The main reason is the blindness of search behavior which leads to the employed bees not generating promising candidate solutions. To address this issue, we propose an improved ABC algorithm (ABCPSE) combined with the previous successful search experience. The proposed algorithm has the following innovative advantages: First, the previous successful search experience with good performances in convergence and distribution are employed in real time to generate offspring. This rule can increase convergence speed and exploitation ability of ABCPSE algorithm. Second, local search is performed near the superior individuals produced in a different generation. Hence, a set of solutions with excellent diversity and convergence is obtained. To assess the performance of ABCPSE algorithm, experiments are conducted on a set of 18 benchmark functions. The results demonstrate that the proposed algorithm can produce higher quality solutions with faster convergence than some current state-of-the-art ABC-based algorithms.