| 摘要 |
A method of automatically isolating item faults is disclosed. Subtest results T[i] are obtained for a plurality of items, where T[] comprises a vector of m subtest results and T[i]=a value if subtest i fails and subtest i does not fail, T[i]=a negative value or 0. The subtest results T[i] are multiplied or ANDed with respective matrix values Y[i,j], where Y[] comprises a predetermined mxn matrix where Y[i,j]>0 if hardware module j causes subtest i to fail and Y[i,j]<=0 if item j does not cause subtest i to fail. The results are summed in order to obtain a number S[j] for each item that is reflective of the number of subtest failures that are explained by a failure of item j, wherein any item j is suspect if S[j]>0. The item that has the most likely failure is determined by obtaining the largest value of S[j] when S[j] has multiple non-zero entries, such that the largest value of S[j] is indicative of the most likely single item failure where any other items with non-zero S[j] represent alternate possible item failures.
|
