Research And Improvement Of Artificial Bee Colony Algorithm

A variety of swarm intelligence optimization algorithms are applied to solve various engineering problems,such as system control,production scheduling,pattern recognition et al.Swarm intelligence optimization algorithm is becoming more and more famous in the scholars.In 2005,artificial bee colony algorithm is designed by Karaboga.ABC is a new type of swarm intelligence optimization algorithm.Because of its advantage,such as less parameters,simple operation,strong ability to develop et al,ABC has been widely used to solve engineering problems.The engineering problems include Data mining,the filter of design,neural network optimization and so on.Excellent optimization performance and perfect application have made ABC become the hot research area in modern times.In this paper,the research results mainly includes three aspects as followed:(1)When the test objects are the high dimensional complex optimization problems,the convergence performance of ABC is poor,especially low convergence accuracy.To solve this problems,an improved ABC algorithm(BAABC)is proposed.In the onlooker phase,BAABC algorithm abandoned the roulette selection strategy,but directly selected the nectar with higher fitness.In BAABC algorithm,the attractor is introduced to the search method of onlooker.All onlooker bees moved toward to the attractor in the same proportion.All onlooker bees together exploited the same area.So the exploitation capacity was enhanced.Experimental results showed that the exploitation capacity of the global search was remarkably enhanced.About the iteration number or iteration time,convergence speed is improved obviously.To solve the high dimensional complex optimization problem,the advantage of BAABC algorithm is obvious.What is more,the convergence performance of BAABC algorithm had nothing to do with the dimension of problem,and the robustness of BAABC is very strong.BAABC do well in solving high dimensional complex optimization problems.(2)To solve the inefficiencies in the process of optimization,a novel artificial bee colony algorithm based on feedback and the law of jungle(LFABC)is proposed.The feedback mechanism is introduced in the global formula by LFABC.This way can help the algorithm to improve the search efficiency,reduce invalid search.LFABC directly search the area where the optimal solution is possible to exist,and does not search all area.Linear differential strategy is added in the search equation to balance the exploitation capacity and exploration capacity in every phase.To void dropping into local optimum,LFABC adopts a novel mutation strategy.LFABC algorithm randomly selects a poor individual to initialize.Experiments prove that the mutation strategy effectively prevents algorithm from falling into the local optimum.The convergence performance of LFABC is superior to ABC.LFABC effectively improves the convergence precision,and its convergence speed is very prominent.(3)To work out the low convergence accuracy of the original algorithm,a novel artificial bee algorithm with self-perturbing(IGABC)is designed.IGABC algorithm improves original ABC algorithm from two aspects.On one hand,IGABC adopts boundary improvement scheme to solve these individuals which cross border.On the other hand,IGABC adopts adaptive search equation with self-perturbing to enhance the exploitation capacity.The test objects select 18 benchmark test function and 6 improved ABC algorithms.The experimental results show that the convergence performance of IGABC algorithm has been greatly improved.Compared with the convergence performance of other six improved ABC algorithms,the convergence accuracy of IGABC is increased by 16 orders of magnitude,when the test objects are Rosenbrock and Schaffer whose optimum is very difficult to be found.