Abstract: The existing Adaptive Sequential Diagnostic algorithm（ASD algorithm） based on PMC fault model can not take full advantage of all of the test results. In  order  to reduce the number of tests and improve diagnostic efficiency, a New Adaptive Sequential Diagnosis algorithm（NASD algorithm） is presented. The concept of relative fault unit is introduced and the diagnostic theorem of fault processor and non-fault processor is put forward and proved . Thus, system diagnosis strategies are as follows：（1）Fault processors are found while non-fault processor is sought. （2）The processors which have been confirmed as faulty will not be used in any tests. （3）Once an non-fault processor is found or the number of fault processors is close to half of the total, the system diagnosis is stopped. Example shows that NASD algorithm is better than other ASD algorithms.

Key words: system-level fault diagnosis, PMC fault model, adaptive diagnosis algorithm, New Adaptive Sequential Diagnosis（NASD） algorithm

摘要： 在PMC故障模型下，现有的自适应顺序诊断算法（ASD算法）不能充分利用所有的测试结果。为了有效地减少测试次数，提高诊断效率，提出一种新的自适应顺序诊断算法（NASD算法）。引入相对故障单元的概念，给出并证明了故障单元和无故障单元的判别定理。据此给出系统诊断的策略：（1）边寻求无故障单元边确诊故障单元;（2）已确认的故障单元不再参与任何测试;（3）找到无故障单元或故障单元数接近一半时，系统诊断结束。实例表明，NASD算法优于其他ASD算法。

关键词: 系统级故障诊断, PMC故障模型, 自适应诊断算法, 新的自适应顺序诊断（NASD）算法