Abstract: There are two main types of algorithms for Chwa & Hakimi fault model，i.e.t-diagnosable algorithm and equation-diagnosis algorithm.However，either of them has its two sides：The former requires the number of fault processors should be less than half of all processors while the latter desires as more of the fault processors as possible.In the paper，a guideline for using t-diag-
nosable or equation-diagnosis algorithm in a specific case is first provided.Then a dichotomizing diagnosis algorithm is established，in which the original test system is divided into two parts，which are relative normal processors set and relative fault processors set，when the number of fault processors is about half of the total of all processors.Hence a self-adaptable algorithm can be used for respective processor set.

Key words: system-level fault diagnosis, Chwa &, Hakimi fault model, t-diagnosable, equation-diagnosis algorithm, dichotomizing diagnosis algorithm

摘要： 在对Chwa & Hakimi故障模型的诊断中，目前相对成熟的算法有t-可诊断性算法和方程诊断算法两大类。然而，上述两类算法各有其优缺：前者要求故障处理机的数目小于处理机总数的一半；后者则希望故障处理机的数目多多亦善。不仅指出何时采用t-可诊断性算法或方程诊断算法，而且建立了所谓的二分诊断算法，即当故障处理机数量占处理机总数一半左右时将原测试系统拆分为两部分：相对正常机集合和相对故障机集合，从而对各个处理机集合采用各自适合的算法去诊断。

关键词: 系统级故障诊断, Chwa &, Hakimi故障模型, t-可诊断性, 方程诊断算法, 二分诊断算法