Remote acoustic detection of cuts in a vibrating plate with stochastic input forcing in a reverberant environment.

Abstract

A mechanical structure subject to vibratory forcing will often radiate sound. When remotely recorded, this sound depends on the vibratory forcing, the structure's frequency response function, and the structure-to-receiver acoustic propagation. Thus, successful remote acoustic detection of changes in a structure's vibration response between baseline and subsequent test recordings may require compensation for possible baseline-to-test changes in vibratory forcing and acoustic propagation. Compensation schemes for unknown structural forcing in an unknown reverberant environment that allow such remote detection are described here and found to be successful when the random forcing has a consistent power spectrum and the structure-radiated sound is recorded with an array of receivers. In particular, experimental results are presented for remote acoustic detection of 13-76 mm cuts in a vibrating 0.30-m-square by 3-mm-thick edge-clamped aluminum plate subject to 0.1-2 kHz base excitation in a reverberant laboratory. Radiated sound from the plate is recorded remotely with a 15-element microphone array and processed with the synthetic time reversal blind deconvolution algorithm to compensate for unknown reverberation. Cut detection success is compared for frequency-sweep and random-input forcing when this forcing is known and unknown, and when there are plate-to-array geometric changes between the baseline and test measurements.