| In recent years, Web service is widely used as a method for realizing the service oriented architecture (SOA). More and more organizations release software and data on the Internet by packaged as Web service. During the process of service used, the provided function of single Web service is often difficult to satisfy the users, then need to compose services together to realize more complex function. Web service composition greatly improved the efficiency of software resource utilization and the efficiency of the development of enterprise application, completed the task which one single Web service cannot complete, and provided a way for the cooperation between enterprises and organizations, to become the key technology of the enterprise application integration.Through the service composition can rapidly, effectively provide new services to users. But with the increasing number of Web services on the Internet, Web services with same function attribute but different non functional properties (i.e. QoS) are also more and more. It is became a hot topic of research that how to choose the Web service with the QoS requirement which could satisfy the user from the candidate service set. Many existed models of QoS use the certain parameters to selecting service. However, in the real network, since there are a large number of uncertain factors, the runtime QoS of the composite services which was chosen by the existed QoS models (using the certain parameters) may become poor, or even cause acceptable. How to select service according to the uncertainty QoS. And making the chosen service can meet the needs of users in most cases becomes a problem which must be solved in the field of service selection.In this paper, based on the study of the research of QoS-based Web service selection, mainly used the workflow model, finished the following studies about uncertain QoS service selection:The research for the QoS driven Web service selection, this paper first studied the supporting technologies of Web service, Web service composition and Web service selection. Second analyzed the service selection problem based on the QoS, established the determine QoS model. Then according to the uncertainty in QoS, established the QoS model with uncertainty, and introduced the uncertain QoS into the service selection model. At the same time, given the pretreatment method of the model.Considered the robustness of the robust optimization method in solving uncertain programming problems, this paper introduced the robust optimization method to solve Web service selection model. Realized the robust optimization for uncertainty QoS based service selection. Based on this, I proposed the splitting method for workflow model, and simplified the selection model, so the calculating time is greatly shortened.Finally, in order to use the method this paper proposed, designed and implemented a service selection and test prototype system based on JSP. In the system, the user can use the service selection method by proposed. And it embedded a automatic test module, which makes the collection of service QoS data is more convenient. |