Abstract: By studying the mission needs and the previously actual collaborative design capacity of Agent，that is，the commitments made during the mission carrying out and the real pay，the pay/commitment relationship to get the reliability degree of the collaborative capacity.Using the ant colony algorithm，which is improved by the cloud model combined with the optimal tree to analyze the reliability degree can restrict the traditional ant colony algorithm trapped into local optimal solution effectively.This improved algorithm has been applied to the personnel selection of collaborative design.Authors find that this algorithm can not only provide decision making support for design manager，but also match the design staff better.Further more，it can make the whole design orderly and stable and can achieve its optimal function.All of these would provide support for the new projects’ carrying out.

Key words: cloud models, Ant Colony Algorithm, computer supported cooperative design, reliability degree

摘要： 通过研究任务需求和Agent曾经的协同设计实际能力，即执行该任务时做出的承诺和实际做出的付出，付出/承诺关系，得到协同能力信度。并利用云模型结合最优树理论改进蚁群算法来分析此信度，有效地限制了传统蚁群算法易陷入局部最优解的问题。把改进的算法应用到协同设计中的人员选择中，发现此算法可以为设计管理人员提供决策支持，可以更好地匹配设计人员，可以实现整个设计的有序、稳定和整体功能效应的最佳，可以为新项目的开展提供支持。

关键词: 云模型, 蚁群算法, 协同设计, 信度