Boiling detection experiments by acoustic analysis applying linear discriminant function

Abstract

The detection of the boiling is examined by acoustic analysis applying the pattern recognition technique. The subcritical assembly was used as a sham reactor vessel. The vapor bubbles at the surface of the stainless steel heater generate the sound when they vanish in the subcooled water. These sound are detected by the hydrophone and then transformed to the power spectral densities(PSD). The theory of linear discriminant function is applied to classify the PSD data. The generation of the boiling is detected by the computer which distingwish the PSD with boiling sound from that without boiling sound. The accuracies of discriminations were very high when the PSD of 2 to 6 KHz range were discriminated in these experiments.