Web Services Composition Optimization Research Based On Improved Artificial Bee Colony Algorithm

With the rapid development of web services,web services composition becomes more and more complex.There have been many web services with similar properties,which non-functional properties are not same.So how to combine Web services is a NP-hard problem.This dissertation proposes an improved artificial bee colony algorithm to slove optimization on web service composition.There is main work as follows:First,this dissertation proposes elite-space strategy to change foraging strategies of lead bees.Through conjunction with genetic algorithms,this model effectively avoids the occurrence of precocious and optimizes the population of web service.The different solutions of multi-objective problems directly exist mutual conflicts.If improving a solution,other solutions in most degree may reduce.In order to make all solutions optimal,the dissertation proposes the concept of non-dominated set.In this situation,multi-objection optimization problems change into a group set that meets Pareto.Generally,the problem of web service optimization will only launch a optimal solution.To meet accuracy of requirements,there will be a set of optimal solutions to optimize population based on Pareto set.Then the dissertation expands the population range by using exchange strategy of utility value on followed bee.Initially,this original artificial bee colony algorithm in the selection of followed bees uses roulette bet strategy.Bees with bigger adapted degrees will be selected.Because of existing the situation of local optimization,the dissertation uses followed bees searching by flexibility-resource model to improve algorithm.The model can expand the scope of the search based on Pareto set.This dissertation also proposes a series of formulas to determine the initial position and location of the new bees for the optimization of web services composition.Finally,through simulation experiments to prove this scheme,this algorithm effectively avoids precociousness,increases the diversities of the population and obtains solutions to meet the users' actual needs.Through experiments,it proves the improvements to the algorithm are true and valid.