Font Size: a A A

Research On Improvement Of Cuckoo Search Algorithm

Posted on:2018-01-31Degree:MasterType:Thesis
Country:ChinaCandidate:C MaFull Text:PDF
GTID:2348330542469776Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
Generally,optimization problems are divided into two categories.One is the continuous optimization problem.The other is the discrete optimization problem.With the development of society and the progress of science and technology,the number and complexity of the optimization problem are increasing.The traditional mathematical method can not meet the requirements of complex optimization problem,so the intelligent algorithm comes into being.Cuckoo search(CS),as a new swarm intelligence search algorithm,has certain advantages in solving the above optimization problem,but the algorithm in search procedure is of slow convergence and poor accuracy.Therefore,researchs on improving Cuckoo Search algorithm and using it to solve single-objective&multi-objective function and combinatorial optimization problems are of important theoretical significance and potential practical value.The main research contents of this article are as follows:(1)Through summarizing the research status of the Cuckoo Search algorithm on the single-objective optimization problem from four aspects:algorithm hybrid,adaptive search,update strategy and discretization,and the research status of the Cuckoo Search algorithm on the multi-objective optimization problem from four aspects:update mechanism,file generation and maintenance,algorithm application,the problems of the Cuckoo Search algorithm on the single-objective and multi-objective optimization problems are discussed respectively.The content of the research is determined,and the research ideas are raveled out.(2)The CS algorithm in search procedure is of slow convergence and low accuracy.To improve convergence speed and optimal accuracy,a Cuckoo Search algorithm combined with simulated annealing(SA-CS)is proposed.Simulated annealing mechanism is introduced to gain a worse solution at a certain probability to ensure the algorithm run beyond the local optimum and enhance the algorithm's ability to find the optimal solution if the solution falls into local optimal value by using the rules of annealing timing.Through testing the benchmark functions,the effectiveness of SA-CS algorithm is verified.Aiming at the problem of solving typical discrete TSP problem,the eliminating path crossing mechanism is introduced to improve the accuracy of the algorithm.The results show that SA-CS provides a better optimization precision and convergence rate for the function optimization problems and the combinatorial optimization problems.(3)The MOCS algorithm can't achieve the uniform,diverse and convergent Pareto front for multi-objective problems.To get the better Pareto front,a novel multi-objective Cuckoo Search algorithm based on decomposition(MOCS/D)is proposed.The Tchebycheff decomposition method is introduced into the MOCS to improve the convergence ability of the algorithm.Through testing the benchmark multi-objective functions,the results show that MOCS/D can get a better convergent,diverse and uniform Pareto front than MOCS for the multi-objective function optimization problems.(4)To overcome the poor performance of MOCS algorithm on the combinatorial optimization problems,a binary multi-objective Cuckoo Search algorithm based on decomposition with uniform design(UMOBCS/D)is proposed in the paper.In the proposed algorithm,the uniform design method is introduced to get the evenly distributed weight vectors on the basis of MOCS/D algorithm.Meanwhile,the binary transformation mechanism is applied to the update process of CS algorithm for the characteristics of 0-1 knapsack problems.Likewise,the repair strategy of infeasible solutions is used to satisfy the constraint conditions.Through testing the 12 multi-objective knapsack problems,the simulation results show that UMOBCS/D algorithm has a better performance than MOCS algorithm and NSGAII algorithm for both the multi-objective knapsack problems and combinatorial optimization problems.
Keywords/Search Tags:Cuckoo Search Algorithm, Evolutionary Algorithm Based on Decomposition, Multi-objective Optimization, Function Optimization Problems, Combinatorial Optimization Problems, Simulated Annealing Algorithm
PDF Full Text Request
Related items