A Comparison of Matched-Field and Pseudo-Inverse Methods for Source Identification on a Frame Structure

The combined use of pseudo-inverse and matched-field processing methods is investigated for localizing and characterizing a broad-band random vibration source on a lightly-damped finite frame structure. The low-frequency range where global modes of the frame are dominant in the response is considered. Experimental results for two different sensor array geometries are compared. Also, the use of numerically-derived transfer functions is demonstrated for each of the processing methods. The results indicate that a combined use of matched-field and pseudo-inverse methods allows for a more complete characterization of the frame excitation than either method used exclusively.