Abstract: The algorithm of services selection is a key factor which can affect the QoS of composite service and the performance of service composition.The problem of Web service selection based on QoS hasn’t been essentially solved by a class of selection algorithms that aggregate multiple QoS parameters to form a composite objective function for optimization using weighted scoring method.In this paper，a Web service selection approach based on an improved strength Pareto evolutionary algorithm（SPEA2+） is designed to solve this problem by multiobjecitve optimization.The Pareto-optimal solutions are obtained by simultaneously minimizing service time（T） and service cost（C） with satisfying the constraints of the parameters reputation（RE），reliability（R） and availability（A）.Some experiments are made to compare SPEA2+ with SPEA2 for solving this problem and the results show both SPEA2+ and SPEA2 have their advantages and disadvantages in some cases，and users can select solutions according to their preferences for T or C.

Key words: Quality of Service（QoS）, Web Service, services selection, SPEA2+

摘要： 服务选择算法是影响组合服务的QoS和服务组合性能高低的关键因素。针对将多维QoS属性加权聚合为一个目标函数进行优化的选择算法不能实质性地解决基于QoS的Web服务选择问题，采用改进的强度Pareto进化算法（SPEA2+）多目标优化求解该问题，设计了一种Web服务选择算法。该算法在满足声誉、可靠性和可利用性属性的约束下，同时最小化服务时间T和服务费用C，以产生Pareto最优解集。通过与运用SPEA2算法求解该问题的实验对比，表明了两算法所获取的Pareto最优解集中QoS属性T和C各均值各有优劣，用户可依据对T或C的偏好择优选择。

关键词: 服务质量, Web服务, 服务选择, SPEA2+