The Algorithm Study Of Fault Diagnosis Based On The PMC Model

In this thesis, we mainly talk about the system-level fault diagnosis algorithm based on the PMC model about the large multi-processors system and the computer network system. Three new diagnosis algorithms are presented here.The probability matrix diagnosis algorithm is entirely based on the calculation of the probability and the matrix. It shows the test result of processors in fault system with the matrix, and then divides all of processors into many quasi bodies by the matrix calculation. According to the probability calculation, the algorithm can find that the probability of which elements in the quasi body are faulty is less than the agreed value (generally 0.001), and then get hold of a non-fault quasi body. It can get other non-fault quasi bodies in accordance with the non-fault quasi body, and ultimately find all of the non-fault quasi bodies in the fault system.The SIDSD algorithm is an improved algorithm. Particularly, it has the advantage in the ways of building and finding the ring which is made up of non-fault processors. When using the algorithm, you can get a diagnosis result quickly after diagnosis.The dichotomy diagnosis algorithm, which is based on the equation diagnosis algorithm and the body algorithm, solves the problem that the calculation of the equation diagnosis algorithm is more complex and the body algorithm is difficult to diagnose when the number of fault processors is very close to the number of non-fault processors.