TDOA-based localization of cracking sound events with minimal-error microphone subsets

We present a method based on the time difference of arrival (TDOA) and index minimal-error subsets of microphones to localize sudden cracking sound events, which appear somewhere in flax-fiber reinforced concrete specimens. Validation tests with small-scale pendulum impacts and known impact locations were carried out. Error estimation was performed and error ellipses were calculated. Microphone subsets leading to the smallest localization error were indexed. We validated the localization accuracy against localization results calculated using the delay-and-sum beamforming technique. Further, tension tests on concrete specimens were performed until failure; crack patterns were recorded by photogrammetry. The cracking sound events were localized. With the good match between TDOA-based localization results and crack patterns, we demonstrate that the proposed localization procedure is reliably applicable for real-time localization of concrete cracking.