Research And Application Of Multi-target Group Search Algorithm

Multi-objective Optimization Problems (MOPs) is a very important problem existing in the field of engineering practice and scientific research. How to get the optimal solutions of the multiple conflicting objectives has always been the focus of academic and engineering circles. In recent years, with the development of swarm intelligence optimization algorithm, the multi-objective genetic algorithm, the multi-objective particle swarm optimization algorithm and many other multi-objective swarm intelligence optimization algorithms have been widely applied in solving MOPs. Group Search Optimization algorithm (GSO) is a new type of intelligence optimization algorithm, which is based on the the birds, lions and other animal foraging behavior and established according to Producer-Scroungermodel. GSO has the superiority in solving the single objective problem and shows a good performance in high-dimensional function optimization. To solve the problem of complicated engineering, the GSO algorithm has obvious advantages. However, as a new type of intelligence optimization algorithm, GSO is less studied in solving the multi-objective optimization problems. Especially for collaborative multi-objective group search algorithm, the research is not many which is worthy of further study.Due to the rapid development speed of the Internet, and the explosively growing of the users, the network ossification phenomenon has been more and more serious. Network virtualization provides a new way to solve this problem. As the main challenge to realize network virtualization, the virtual network embedding problem has been proved to be a typical NP - hard problem. For this reason, a new kind of virtual network embedding based on multi-objective group search optimizer is proposed in this paper, which can prove the practical significance of the multi-objective group search algorithm and provides a new method for solving the virtual network mapping problem.Based on the standard group search algorithm and multi-objective evolutionary theory, in this paper, the multi-objective group search algorithm and its application are studied and the experimental data are analyzed. The concrete content is as follows.(1) A novel multi-objective group search optimizer based on multiple producers and crossover operator of genetic algorithm was proposed. The algorithm extends producer to multiple, which could enhance the convergence rate of algorithm and the diversity of non-dominated set.The metropolis rule is intruduced into the search pattern of producers to prevent a local optima solution. At last, to enhance the algorithm's ability to find new solutions and expand the range of non-dominated set. The algorithm combines the crossover operator and rangers' search strategies. Experimental results demonstrate that the algorithm can effectively and efficiently solve multi-objective optimization problems(2) An algorithm of Cooperative Coevolutionary Multi-objective Group Search Optimizer is proposed. In the algorithm, the MOPs are decomposed according to their decision variables and optimized by corresponding subswarms respectively. Collaborators are selected randomly from archive and employed to construct context vector in order to evaluate the members in sub-groups. Finally, combining the members of each subgroup multi-objective Pareto solution set to get the solutions.(3) Multi-objective Group Search Optimizer is used to solve virtual network mapping problem.Virtual network mapping model is established and multiple mapping objective function are constructed to solve the problem of the herogeneous resource metrics hard to unify, and then using multi-objective GSO algorithm to optimize.