中文摘要：

由于难以恰当地描述多个任务两两之间的关系，耦合任务群的分解仍然是一个难题。根据研发团队的结构特点提出两种耦合任务群分解方法：基于敏感度、可变度和任务量的分解方法将布尔型DSM转化成含有敏感度、可变度和任务量信息的矩阵，然后将该矩阵转化成相互作用强度距离矩阵，再进行聚类，使相互作用较强的任务聚在一起；基于任务属性灰度关联相似度的分解方法用灰度分析法计算各任务属性的权重，然后根据任务属性及其权重建立任务距离矩阵，再进行聚类分解，使相似的任务聚在一起。用MATLABWebServer工具开发了基于WEB的耦合任务群分解仿真模块，采用仿真比较的方法分析算法对任务分解结果的影响。

英文摘要：

For the difficulty of relationship describing among tasks, coupling tasks is still hard to decompose. According to structural characteristic of R＆D team, two decomposing methods of coupling task group were put forward. Sensitivity, Changeability and Work volume based on Decomposition （SCWD） translates Boolean DSM into a matrix which has sensitivity, Changeability and Work volume information of tasks, then further translates such matrix into a distance matrix of coupling intensity, and then clusters strongly coupling tasks together. Grey Relational Similitude of Task Attribute based on Clustering （GRSTAC） calculates weight of task attributes by grey theory, then constructs a distance matrix according to attributes value and their weight, and then clusters similar tasks together. A simulation system was established by MATLABWebServer tool. Simulating and comparing method is used to analyze effect of different arithmetic to decomposing results.