The Improvements Of Artificial Bee Colony Algorithm

Biologically inspired computing is a simulation of natural biological, ecologicalsystem,"the survival of the fittest" behavior of the model and algorithm, withadaptive, self-organizing, self-learning etc., it can solve the complex problems betterthan the method of traditional calculation. Bio-inspired computing has shown that itsexcellent performance and great potential for development by solved many complexoptimization problems. In recent years, it has attracted many domestic and foreignscholars have studied many aspects on it.Artificial bee colony algorithm is an intelligent simulation calculation method ofbionic bee colony proposed by Turkey scholar Karaboga in search of good nectarsource. The algorithms have been proposed so far only nine years. Artificial beecolony algorithm has obvious advantages, such as simple principle, less controlparameters, easy to implement, but the convergence speed is very high. It has beenproved to be an excellent global optimization algorithm, there is the concern of manyscholars at home and abroad. However, the research and application of artificial beecolony algorithm at present is still in the primary stage, there are still some problems,such as premature convergence, slowing down of the evolution of post-optimizationand so on.To improve the shortcomings of ABC algorithm, The ABC algorithm theory, thealgorithm framework model and the model of information exchage are also deeplystudied in this paper. Based on these theoretical investigations, this paper presents twoimproved algorithms of ABC. The experiments show that the improved algorithm isfeasible and effective. The main contributions of this dissertation are listed below.First of all, artificial bee colony algorithm in the bee scout stage search operationcan solve the algorithm into local optimal solution in a certain extent, but also likeother heuristic optimization algorithm, there is a poor local search ability, nearoptimal solution search efficiency, is likely to fall into local optimum and make thealgorithm stagnation problems and solving complex problems. In order to improvethis defect, this paper uses NM algorithm to replace the artificial bee colony algorithmscout bee stage of randomly generated individual mechanism, proposes an improvedartificial bee colony algorithm based on NM (NMABC). Hoping NM algorithm basedon excellent local search capability, artificial bee colony algorithm to improve thelocal search ability of poor defect and improve search efficiency. Secondly, considering the mechanism of NM algorithm may lead to repeatedsearch time increased, so this paper also introduced the PT disturbance mechanisminstead of the NM algorithm, proposed a PT disturbance mechanism based onimproved artificial bee colony algorithm (PTABC).Hope to reduce the fitnesscomputation times, to avoid falling into local optimal conditions, improve thealgorithm speed of evolution, so as to better achieve the global optimization and localoptimization balance.Finally, simulation experiment of artificial bee colony algorithm is proposed inthis paper. The experimental results show that, this research proposed two improvedalgorithm to a certain extent, to avoid the artificial bee colony algorithm is easilytrapped into local optimal and precocity.