A FAST ALGORITHM FOR FAULT ISOLATION IN DICTIONARY APPROACH OF ANALOG CIRCUIT DIAGNOSIS

Abstract

Dictionary generation, selection of test nodes and fault isolation are three important phases of dictionary approach of analog fault diagnosis. At present, fast algorithms exist in the literature for the first two phases whereas it is not so for the fault isolation phase. This gap is filled in this paper. First, complexity analysis of existing methods is carried out. The concept of hashing is then applied to achieve fault isolation in the fastest way. In terms of time complexity this is only O(1) as against O( log 2 f) in the best known method, where f is the number of faulty conditions. Thus the computational requirement in the new method is independent of dictionary size. A collision free hashing function useful for this application is also devised.