Artificial Bee Colony Algorithms Based On Self-adaptive Differential Evolution And Across-neighborhood Biogeograpical Migration

Artificial bee colony algorithm(ABC)is one of the most popular swarm intelligence algorithms with less control parameters and sim-ple structure,which is more competitive than other population-based algorithms.However,the algorithm still has some shortcomings in convergence,for example it is good at global exploration,but its lo-cal exploitation ability is very weak,especially when the dimension of the problem increases,the optimization accuracy of the algorithm will be very low.This paper aims at dealing with the difficulty of balancing the global exploration and local exploitation,two improved artificial bee colony algorithms based on self-adaptive differential evo-lution and across-neighborhood biogeography are proposed.Algorithm 1 is an artificial bee colony algorithm based on self-adaptive differential evolution(SADEABC).Inspired by adaptive dif-ferential evolution algorithm using mutation strategies pool to select mutation strategy and corresponding parameters,algorithm 1 intro-duces crossover rate and inertia weight in the artificial bee colony algorithm,and enhances the search ability of the employed bees with differential vector and global best information,which can balance the global exploration and local exploitation.Algorithm 1 utilize differ-ential strategies pool to make onlooker bees choose update strategy independently,and allocate more computing resources to effective s-trategy.In addition,algorithm 1 uses chaotic systems and opposition-based to initialize the population,and more uniformly distributions of the initial solutions are obtained.The SADEABC algorithm is tested on 18 benchmark functions and compared with the five algo-rithms.Experimental results show that the improved algorithm has both convergence speed and solution accuracy with significant com-petitiveness.Algorithm 2 is an artificial bee colony algorithm based on across-neighborhood search and biogeography migration(ANSBMABC).In-spired by the across-neighborhood search algorithm,algorithm 2 adds self-adaptive strategy in the employed bee phase,which can make the honey sources select cross-neighborhood search degrees adaptive-ly and integrate information from multiple honey sources during the update process to obtain the potential components of good solutions.All of this can enhance the global search capabilities of the algorith-m.At the same time,the use of biogeography migration strategy to guide the search of onlooker bees can strengthen the exchange of information between individuals and improve the performance of the algorithm.Numerical experiments are performed on 18 unimodal,multimodal,and rotated benchmark functions,and ANSBMABC al-gorithm is compared with the state-of-the-art algorithms.The results show that the new algorithm can be outstanding in the convergence speed and the optimization accuracy,and the accuracy of solutions is more stable when solving high-dimensional optimization problems.