Implementing statistical analysis in multi-channel acoustic impact-echo testing of concrete bridge decks: Determining thresholds for delamination detection

作     者：Lorin Hendricks W. Spencer Guthrie Brian Mazzeo 

作者机构：1Department of Electrical and Computer Engineering Brigham Young University Provo Utah USA 2Department of Civil and Environmental Engineering Brigham Young University Provo Utah USA 

出 版 物：《AIP Conference Proceedings》 

年 卷 期：2018年第1949卷第1期

学科分类：07[理学] 0702[理学-物理学] 

摘      要：An automated acoustic impact-echo testing device with seven channels has been developed for faster surveying of bridge decks. Due to potential variations in bridge deck overlay thickness, varying conditions between testing passes, and occasional imprecise equipment calibrations, a method that can account for variations in deck properties and testing conditions was necessary to correctly interpret the acoustic data. A new methodology involving statistical analyses was therefore developed. After acoustic impact-echo data are collected and analyzed, the results are normalized by the median for each channel, a Gaussian distribution is fit to the histogram of the data, and the Kullback-Leibler divergence test or Otsu’s method is then used to determine the optimum threshold for differentiating between intact and delaminated concrete. The new methodology was successfully applied to individual channels of previously unusable acoustic impact-echo data obtained from a three-lane interstate bridge deck surfaced with a polymer overlay, and the resulting delamination map compared very favorably with the results of a manual deck sounding survey.