Transforming Sparse Piezoelectric Measurements into High-Resolution Damage Maps for Structural Health Monitoring

Abstract

Structural health monitoring of plate-like structures often relies on guided wave propagation to detect and locate damage. While full-field vibrometric measurements offer exceptional resolution and detail, their high cost and complexity make them unsuitable for widespread use in practical applications. In contrast, piezoceramic transducers provide a cost-effective alternative but are limited to collecting data at discrete points, leading to significant challenges in achieving high-resolution damage visualization. This paper addresses this gap by introducing a novel method to transform sparse piezoelectric measurements into high-resolution visual damage maps. This paper presents a novel, image-processing-based method to reconstruct highresolution damage maps from sparse piezoelectric measurements. The approach leverages emitter-receiver pairs to draw color-coded propagation paths, where intensity reflects the severity of damage detected along the wave trajectory. Unlike existing methods such as RAPID, the proposed framework is computationally lightweight, baselineagnostic in terms of damage indices, and compatible with a wide range of signal features beyond Pearson correlation. The proposed method utilizes an image-processing-based algorithm to reconstruct dense feature maps of damage presence. The approach leverages emitter-receiver pairs to draw color-coded propagation paths, where intensity reflects the severity of damage detected along the wave trajectory. The resulting visualizations closely mimic those obtained from full-field vibrometry, while significantly reducing data requirements and computational overhead. To validate the method, experiments were conducted on an aluminum plate equipped with 8 signal transmitters and a laser vibrometer to measure wave propagation. Seven damage scenarios were tested, ranging from barely visible damage to a 17 mm flatbottom hole. This proof-of-concept offers a practical and scalable solution for highresolution SHM and serves as a foundation for more detailed quantitative studies presented in future work.