Font Size: a A A

Improved Flower Pollination Algorithm And Its Application Research

Posted on:2019-10-27Degree:MasterType:Thesis
Country:ChinaCandidate:L LiuFull Text:PDF
Abstract/Summary:PDF Full Text Request
The basic flower pollination algorithm is a new type of meta heuristic algorithm proposed by Cambridge University scholar Xin-She Yang in 2012,inspired by the mechanism of flower pollination in nature.Once the algorithm is proposed,which was highly concerned and researched by many scholars at home and abroad.The algorithm simulates the behavior of self-pollination and cross-pollination in flower pollination,which corresponds to the local search mechanism and the global search mechanism respectively.The relationship between local search and global search is balanced by using switch probability.Promoting the algorithm to solve the optimization problem through continuous iterations.The algorithm is simple and easy to operate,and good at searching,and has high stability and robust.At present,many scholars have given a series of improvement strategies and researches,which improved the optimization performance of the algorithm,had better convergence accuracy,improved and enriched the overall framework of the basic algorithm and search mechanism in the algorithm.When the improved algorithm was successfully applied to the solving of optimization problems in various fields of actual production and life,good results were obtained.But there are some defects of the algorithm,for example,parameter selection in algorithm lacks proper design basis.When running to the later stage of the algorithm,it is easy to fall into the local extremum and it is hard to jump out,and it is unable to find a better accuracy of solution.All of these directly affect the application scope of algorithm in engineering problems.This paper mainly analyzes and studies the defects of basic flower pollination algorithm.According to different improvement strategies and ideas to improve the algorithm.By enhancing the parameters,framework processes,and internal search mechanisms involved in the algorithm,the overall search ability of the algorithm was enhanced,and made the convergence rate and optimization accuracy of the algorithm improve at the same time.When the improved algorithm is successfully applied to the optimization problem of task resource scheduling under the cloud computing platform,good application results are obtained.At the same time,the scope of application of this algorithm has been expanded accordingly.The works of this article can be summed up in two aspects:(1)An improved adaptive flower pollination algorithm was proposed based on simulated annealing temperature reduction mechanism.Aiming at the pollination algorithm is easy to fall into the local optimum,in this paper,the Levy flight of global pollination used the scaling factor to control step size with deformation index function in the basic algorithm,it made the update of the individual position of the flowers' next generation effectively improve the adaptability with the number of iterations increases.Aiming at the low accuracy of the algorithm,this paper improved the influence factor of reproduction probability by Rayleigh distribution function combining with the iterations.Then the premature convergence of algorithm was avoid,and at the same time,it was more closer to the optimal solution in the later stage.Combined the cooling operation in the simulated annealing algorithm into the improved flower pollination algorithm.Enhanced the overall optimization performance of the algorithm,and enriched the diversity of the population.The data obtained through the simulation test functions and the convergence figures indicates that the convergence accuracy and rate of the improved algorithm have improved by a certain margin.The algorithm has obvious improvement and has certain advantages.(2)An improved flower pollination algorithm combining differential evolution and dynamic transition probability is proposed.On the other hand,according to the improvement strategies based on different thoughts.The switch probability is improved through Weibull distribution function combined with iteration times,the relationship between global pollination and local pollination was better balanced,and the overall optimization performance of the algorithm was improved effectively.The random mutation operator was introduced to the global pollination stage of algorithm,which increased the diversity of population,and improved the optimization ability of the algorithm in the global search,also avoided the algorithm falling into premature convergence.In the process of local pollination,the directional mutation and crossover operation of differential evolution were integrated,which renewed the individual position of flower with memory function.It was reasonable to choose the direction of mutation,meanwhile,the cross operation was used to avoid the new solution beyond the boundary and increase the convergence rate,which made the algorithm continuously approach the global optimal solution.(3)Successfully applied MCFPA algorithm to the optimization problem of resource scheduling under cloud computing platform.Using flower individual location for virtual machine location.Based on the comparison result of the improved dynamic transition probability and the golden ratio coefficient,each flower individual chooses to perform the global search or the partial search.In order to determine the reasonable choice with the best performance of virtual resources for the next task.By defining the time cost and expense of cloud computing resource scheduling to compute the objective function of resource constrains.When the flower individual completed the target search task,the choice of the virtual machine resource corresponding to the individual location of the flower is the best allocation plan for the problem.Comparison of simulation results show that the improved algorithm can save task scheduling costs and time,and have better results than basic flower pollination algorithms.
Keywords/Search Tags:Flower pollination algorithm, Simulated annealing operation, Random mutation operator, Differential evolution, Cloud computing resource scheduling
PDF Full Text Request
Related items