文中介绍了一种系统级故障诊断模型——通用比较模型 .该模型允许处理器作为自身的比较器 ,综合了经典的 PMC模型和 Maeng/ Malek模型的优点 .基于该模型 ,分析了多处理器系统的 t- 可诊断性问题 ,给出了 t- 可诊断系统的特征化 ,证明了一个系统成为 t- 可诊断系统的新的充分必要条件 .其次 ,证明了在通用比较模型中 ,确定故障处理器集的问题等价于求解一个超图的最小横切集 (Minimum traversal) ,该超图是根据多处理器的通信图和比较图求得的 .最后 ,给出了特定情况下的一个系统级故障诊断并行算法 ,该算法是完全分布的 ,并且在信息传输次数方面是最优的 This paper presents a system-level fault diagnosis model - a universal comparison model that allows the processor as its own comparator, combines the advantages of the classic PMC model and Maeng / Malek model.Based on the model, the multi-processor It is proved that a system is a new necessary and sufficient condition for t-diagnose system.Secondly, it is proved that in the general comparison model, the fault is identified The problem of processor sets is equivalent to finding the minimum traversal of a hypergraph based on the multiprocessor communication graph and comparison graph.Finally, System-level troubleshooting parallel algorithm, which is completely distributed and optimal in the number of information transmissions