Improvement Of Search Strategy Of Artificial Bee Colony Algorithm

With the improvement of living standard,the diversification of demand and the rapid development of world science and technology,the optimization problems become more complex in the fields of scientific research and engineering application,and traditional simple algorithms can not solve these optimization problems.So how to design effective algorithms to solve them has always been a favorite research field of domestic and foreign scholars.And intelligent optimization algorithm has good performance in solving complex problems such as multi-objective,high dimensional,non-differentiable and discontinuous,which has attracted a lot of attention from researchers.Artificial bee colony algorithm(ABC)is a new intelligent optimization algorithm which is inspired by the process of honey bee process in the natural world.The algorithm has the characteristics of simple structure,few parameters and easy to implement,which has been highly sought by researchers since it was proposed.But similar to other intelligent algorithms,the traditional artificial swarm algorithm has the problem of slow convergence speed,insufficient mining capacity and easy to get into local optimal that cannot find the global best solution.Aiming at these problems,the major purpose of this thesis is to improve the convergence rate,balance mining and exploration ability,avoid falling into local optimum.The main contributions of the dissertation are summarized as follows:(1)The fundamental idea,principle and procedure of the algorithm are reviewed.The improvements of artificial swarm algorithm in search strategy,parameter setting and initialization phase are described and the applications of ABC in various fields were presented.(2)In order to solve the problem of slow convergence speed,unbalance of exploration and exploitation,and easy to get into local optimal problem,Gaussian distribution weighted search strategy with current optimal solution guidance and random sampling is introduced to improve the mining ability of the algorithm and avoid getting into local optimum in the onlooker stage.Furthermore,the parameters of control dimensional change are introduced so that each generation in the population can not only change one dimension,but also accelerate the convergence speed.Gaussian equation which contains the effective information of the dissolving information is introduced to generate the new solution.The improved algorithm is used for numerical test and the experimental results show that the improved algorithm compared with the standard ABC has better quality,precision,stability and quickly convergence speed,and compared with GBABC,the performance of the algorithm has certain improvement.(3)For the basic artificial swarm algorithm has not enough mining capacity,and any single algorithm can't solve all the problems,a hybrid algorithm based on artificial swarm algorithm and differential evolution algorithm is proposed.The algorithm uses the global optimal individual to guide the update the direction of the candidate to accelerate the convergence speed.Using the algorithm selection strategy based on the information of the candidate solution whether or not successfully enter the next generation to combine with the two algorithms,avoid falling into local optimal and accelerating convergence rate.The hybrid algorithm is simulated,the experimental results show that the algorithm has a significant advantage in dealing with the problem of single peak function,and shows some advantages in dealing with complex functions.