Application Research Of Multi-objective Particle Swarm Algorithm In Web Service Somposition

In recent years, Web Service has been widespread concern as a new technique. The Interface Definition Language WSDL and content delivery format SOAP in Web Services has become a W3C draft and proposed standard. However, a single web service is often unable to meet the needs of complex application in practical applications. Therefore, how to integrate the formation of a single web service more powerful portfolio of services to meet the different users of complex applications have been referred to as a new research hot spot. With Web Service growing rapidly, it is a key problem how to select web service in compositing multiple simple web services to a complex web service. The thesis mainly focuses on studying key technologies of web service composition related to multi-objective particle swarm optimization. The main work includes:Firstly, this paper proposes an improved multi-objective particle swarm optimization. With the study and research on multi-objective particle swarm optimization, the low efficiency and the early-maturing of simple multi-objective particle swarm optimization are discovered. So this paper improves some aspects of multi-objective particle swarm optimization used to deal with the low efficiency and the early-maturing of simple multi-objective particle swarm optimization, such as the retention strategy of Pareto optimal solution and the global optimum selection of particle. In respect of Pareto optimal solution of the reservation, this paper introductions auxiliary container and temporary container to achieve the reservation of Pareto optimal solution, and removes extra particles in the temporary container through the intensive distance. In respect of selection of the global best particle, this paper proposes a new method of selecting the global best particle through the intensive combination of distance and Euclidean distance.Secondly, this Paper tests the improved multi-objective particle swarm optimization in two aspects of convergence and diversity, and compares the test results with other algorithms with tree typical test function. The test results show that the improved multi-objective particle swarm algorithm have good performance in two aspects of convergence and diversity. This paper compares the Pareto curve of test functions received by the improved multi-objective particle swarm algorithm with the theoretical Pareto curve. Further, this shows that the improved algorithm has good performance.Thirdly, this paper applies IDMPSO algorithm to the problem of QoS-based Web service composition, and compares with CMPSO algorithms to the specific Web service composition. The experimental results show that IDMPSO algorithm needs slightly longer running time than CMPSO algorithm, but it can get better solutions than CMPSO algorithm. This show that the feasibility and effectiveness of IDMPSO algorithm.