Leak noise detection in steam generator of liquid metal fast breeder reactors using new statistical features

In a recent communication, the newly defined statistical features were tested for detecting sodium boiling noise and Steam Generators (SGs) leak noise of Liquid Metal Fast Breeder Reactors (LMFBRs) under poor signal-to-noise (S/N) ratio conditions. It was established that the new features, namely the Determinant and Trace of the covariance matrix derived from the appropriately selected components of the Auto Power Spectral Density (APSD) of noise signals, are more sensitive than the conventional features and detect sodium boiling noise with better discrimination for data with an S/N ratio up to - 12 dB and for SG leak noise data with S/N ratio of - 6.7 dB. The sensitivity of these newly defined features has been tested further on leak noise data with S/N ratio ranging from - 8 to - 22 dB. The present paper provides the results and details of the application of the new features on SG leak noise data with S/N up to - 22 dB. (author)