Abstract: To solve the optimization problem that a Service Network(SN)was customized to fulfill a personalized customer requirement in a cost-effective way,a heuristic algorithm was proposed.Two customization mechanisms of multi-source parameters and compound service nodes were used to realize customizable,and there might be multiple customization solutions for a specific requirement.Thus the optimal customization mechanism could be selected.When SN failure to satisfy the requirement,a solution that nearest to the expected requirement should be found.A heuristics algorithm was adopted to find the minimum number of input parameters to be complemented,and an Artificial Bee Colony(ABC)-algorithm was put forward to find a customization solution which was closest to the expectation.A set of experiments showed that the customization performance and efficiency were affected by strictness of requirements and number of multi-source parameters in the SN significantly,and the number of compound service nodes in the SN had comparatively less influence.

Key words: service network, personalized requirements, service mass customization, competency defects, artificial bee colony algorithm

摘要： 为求解服务网络以成本有效的方式满足客户个性化需求的优化定制问题,提出一种启发式定制方法。服务网络通过多源参数和复合服务节点两种机制实现可定制性,针对特定个性化需求可有多种定制方案,需从中选取最优性能的定制方案。当服务网络因能力不足而无法完全满足需求时,找出与需求期望最接近的定制方案。在该方法中,采用启发式方法发现需求无法满足时最少需要补充的输入参数,采用人工蜂群算法发现与期望的服务质量距离最小的定制方案。实验发现,需求的严格程度、服务网络中包含的多源参数数目和服务节点数目对定制效果和效率具有较大影响,而复合服务节点数目对定制效果和效率的影响较小。

关键词: 服务网络, 个性化需求, 服务大规模定制, 能力缺陷, 人工蜂群算法