Abstract
Acoustic sensing technique provides an efficient measure to detect and diagnose incipient failures occurred in plant components. The study reported here investigates the feasibility of adaptive super-directive microphone array technique for detection and localization of anomaly in nuclear power plants. The technique extracts anomaly information of objective components from the acoustic signals obtained at geometrically arranged multiple microphones. A specific signal processing was used to obtain a super-directive sensitivity of the microphone array and to adapt the sensitivity following the acoustic environmental change. Two adaptive sensing algorithms have been developed and implemented in a personal computer, and their abilities to extract the objective acoustic signal have been tested through numerical simulation. For appropriate number of the microphones, the satisfactory performance of the sound extraction has been obtained within reasonable computational load. The practical applicability of this technique to component monitoring in nuclear power plants has been confirmed through the present study.
