Effects of Hologram Distance and Regularization Techniques on Various Methods of Nearfield Acoustic Holography Applied to Building Leakage Detection/Quantification

摘要

This paper attempts to assess the performance of various regularization methods applied to various approaches of nearfield acoustic holography (NAH) in the context of building leakage detection and quantification. Current study addresses one of the important challenges in real building tests, namely, the hologram distance. For a fixed grid resolution, as the hologram distance increases, the error in the reconstruction also increases. This problem can be managed by using a proper regularization method. Various regularization methods are compared for various approaches of NAH. From the current results, Tikhonov regularization technique in combination with a parameter choice method called generalized cross validation (GCV) has the potential to be used as a real life tool for building leakage detection and quantification. The effects of sensor phase mismatch error and signal quantization error are also discussed. This study was performed on a wooden building model that houses a synthetic acoustic source. Microphone measurements were performed on a two dimensional plane near the surface of the building model. Various methods of NAH were applied on the measured data to reconstruct the sound pressure levels (SPL) on the building model surface.