MULTI-FREQUENCY APPROACH TO FAULT DICTIONARY OF LINEAR ANALOG FAULT DIAGNOSIS

Abstract

An efficient method to eliminate redundant frequencies present in one of the existing multi-frequency methods for analog fault diagnosis is proposed in this paper. First the two-dimensional fault dictionary is constructed where entries are gain signatures of all faults and frequencies. The faults belonging to the same quantization levels are numbered sequentially and a frequency that has an ambiguity set with the highest number faults is eliminated after verifying that there are no repetitions after the deletion of this frequency. In this manner, all frequencies are examined for deletion. Finally the test frequencies, which cannot be deleted, remain resulting in a minimal set of test frequencies of a network to isolate a given set of faults. Another method proposes a technique, which generates more number of frequencies to isolate all the faults, if the test frequencies generated using the existing methods are not sufficient.