Application Research And Improvement Of Cuckoo Search Algorithm

With the continuous development of human society and the advancing of the pace of scientific exploration, people come across more and more optimization problems. Due to the increasingly extreme human pursuit of efficiency, conventional optimization methods have been difficult to solve the increasingly complex optimization problems within the limited time and space. Bionic swarm intelligence optimization algorithms become a common and effective way to solve the optimization problems because of theirs flexibility and efficiency.Cuckoo search algorithm is a new heuristic bionic swarm intelligence optimization algorithm; it can be widely applied to various optimization problems because it has some advantages, such as clear theoretical definition, less parameters, easy to be expanded, strong global search ability, easy to be implemented and so on. This paper mainly studies the application researches and improvements of cuckoo search algorithm.This paper introduces the biological principles and mathematical principles of cuckoo search algorithm and introduces Levy flights mechanisms, then expounds the concrete steps and implementation process of the cuckoo search algorithm. On the basis of analyzing the defects of basic cuckoo search algorithm, the basic cuckoo search algorithm is improved, this paper proposes a cuckoo search algorithm based on the Boltzmann selection mechanism, aims to improve the cuckoo search algorithm’s performance in the function optimization problems, make cuckoo search algorithm have higher efficiency of solving problems, faster convergence speed and better solution precision.According to the characteristics of the cuckoo search algorithm, this paper discretizes the cuckoo search algorithm, and put forward a discrete cuckoo search algorithm combined with the idea of tabu search, then applies the discrete cuckoo search algorithm to a classic combinatorial optimization problem-the traveling salesman problem.Cuckoo search algorithm has parallel characteristic in essence, and it is fit for parallel computing. This paper discusses a parallel approach of cuckoo search algorithm, and proposes a parallel approach of cuckoo search in the MapReduce parallel programming model on Hadoop, cuckoo search algorithm is abstracted into several components which are corresponding to several interfaces based on MapReduce, thereby providing a scalable solution for a class of cuckoo search algorithm parallelization based on MapReduce parallel programming model. Finally, simulation results validate the correctness and effectiveness of the parallel approach.Finally, this paper summarizes the whole research content and gives some work prospects in future.