Applying signal statistical analysis to TEOAE measurements

Abstract

Automated detection of TEOAE gains increasing importance in screening applications. Conventional signal statistical analysis methods (buffer correlation and ± difference) of TEOAE waveforms are compared with one another and with a method based on binomial statistics. It turns out that binomial statistics are more reliable and valid than conventional methods with common criteria. In particular, under narrowband noise conditions they are also faster. A frequency specific evaluation is hardly possible by means of conventional methods because of the high random correlation of narrow band signals.