Improved Atificial Bee Colony Algorithm

With the rapid development of science and technology,problems in scientific and engineering fields have become more and more complex.For its large amount of computation,many complex optimization problems cannot be effectively solved by the traditional optimization methods.Therefore,how to design an efficient algorithm to solve the problems in various fields is essential in scientific and engineering applications.Recently,evolutionary algorithms(EAs)have shown considerable success in solving optimization problems with discontinuous,non-differentiable,and multimodal characteristics,which has attracted increasing attention in recent years.At present,EAs have been widely applied in many fields.Artificial bee colony algorithm(ABC)is a hot research topic of EAs in recent years.ABC is based on simulating the foraging behavior of the honeybee swarm.Due to its simplicity,ease of implementation and few parameters,ABC has captured much attention since its invention.However,there are still many issues that need to be investigated and resolved.The major purpose of this thesis is to improve the convergence rate,exploitation ability and solution accuracy of ABC.The main contributions of the dissertation are summarized as follows:(1)The fundamental idea,principle,operators and computing process of ABC were reviewed.In addition,the advantages and disadvantages of ABC were analyzed and the successful applications of ABC in various fields were presented.Moreover,the issues of ABC to be used in engineering applications and the current research directions of ABC were summarized.Finally,the underlying ideas and deficiencies of several typical improved ABC were introduced.(2)An improved ABC with Rosenbrock search and directional information(RDABC)was proposed.Aiming at the slow convergence rate and weak exploitation ability of the basic ABC,RDABC employs the directional information to guide the search process towards the promising regions.Moreover,the best solution in the current population is undergone the Rosenbrock search method to accelerate the convergence rate.The performance of the proposed RDABC was evaluated on eight well-known benchmark test functions.The experimental results demonstrated that the proposed RDABC can achieve better solutions and faster convergence rate than the traditional ABC algorithms.(3)A Rosenbrock ABC with elite region learning(ERABC)was proposed in this study.In the proposed ERABC,it utilizes an enhanced search strategy with elite region learning to maintain the population diversity.Moreover,the Rosenbrock's rotational direction method is employed to improve the exploitation ability.The proposed ERABC was tested on 20 benchmark functions including unimodal,multimodal,and shifted functions.The effects of the improved strategy in ERABC were experimentally investigated.Furthermore,ERABC was compared with some state-of-the-art ABC variants and several related evolutionary algorithms.The experimental results indicated that ERABC enhanced the convergence speed and exploitation ability.