Probabilistic diagnosis algorithm is very important in the system level fault diagnosis research.This paper proposes four probabilistic algorithms based on grouping theory and greedy principle for system-level fault diagnosis.By computer simulation

it is shown that these diagnosis algorithms can achieve a high probability of correctness under low time complexity.The greedy algorithm one has the best performance in the four probabilistic algorithms.The results also indicate that our algorithms have better performance than the Compete algorithm and Majority algorithm

which are classic probabilistic algorithms in system level fault diagnosis.