CORRELATION-BASED COMPARISON OF ANALOG SIGNATURES FOR IDENTIFICATION AND FAULT-DIAGNOSIS

Automatic fault diagnosis and circuit identification using analog sign atures are of great practical interest. The problem of automatic compa rison of nearly similar analog signatures is addressed in this paper. Two known correlation-based similarity criteria are compared, and a ne w criterion is introduced by the use of a weighting factor, which impr oves the discrimination capability. Comparative results for various si gnatures are given, and useful applications of this work are discussed .