Multichannel spectral analysis for tube leak detection

Research to develop multichannel spectral analysis methods, including classical and parametric techniques, for tube leak detection in fossil-fueled power plant boilers is described. Acoustic analysis is employed to detect leaks. The main idea is to monitor the continuous noise environment of the furnace and to look for specific acoustic signatures produced by leaks. A signal spectrum is estimated using the LabVIEW software package. The coherence spectrum of 2-channel signals estimated by the autoregressive technique is shown.<
>