A parametrically optimized multichannel impact-echo system for void detection in tunnel linings

This is the authors' abstract. We don't add key points for this paper.

To address the trade-off between defect-detection reliability and spatial sampling efficiency in impact-echo (IE) assessment of tunnel linings, this study investigates impact-to-acquisition (I-A) spacing as a configurable parameter and examines its effects on spectral characteristics and defect detectability. Experiments on a full-scale C40 curved slab with eight embedded void simulants (10–25 cm side length and 10–30 cm burial depth) show that increasing I-A spacing reduces the true positive rate (TPR) and increases the false positive rate (FPR), while reducing spatial sampling density. A weighted efficiency index (WEI) combines TPR, FPR, and a normalized efficiency factor. Among the six tested spacings and under the initial weights, 14.1 cm produced the highest WEI (0.578; TPR = 0.676; FPR = 0.450), indicating a screening-oriented compromise, whereas 7.1 cm was favored when defect detectability received greater weight. These values were implemented in a four-channel system with interchangeable square-array brackets and a two-stage workflow comprising coarse screening and targeted reinspection. In the laboratory comparison, the two-stage heat map reproduced the preset anomaly locations qualitatively using 156 rather than 960 reported impact locations, corresponding to an approximately 84% reduction in spatial impact locations; quantitative equivalence in detection accuracy was not evaluated. A field demonstration established operational deployability and the acquisition of interpretable thickness-frequency responses, but not field defect-detection accuracy because no independent ground truth was available. The results provide a conditional, parameter-based framework for selecting IE configurations within the investigated specimen, defect, and weighting conditions. Continuous weight-space analysis over the complete admissible domain showed no universally preferred spacing: 14.1 cm occupied the largest conditional preference region and was selected for screening, whereas 7.1 cm was favored in reliability-oriented regions and was selected for targeted reinspection.