Localization of underground pipeline intrusion sources using cross-correlation CNN: application in pile-driving model test

Abstract

Preserving the structural integrity of critical infrastructure systems necessitates a heightened focus on fortifying the protection of underground pipelines. To this end, this paper presents an innovative approach, namely the Multi-Sample Joint Localization Method (MSJLM) utilizing Cross-Correlation Convolutional Neural Networks (CC-CNN), aimed at precisely localizing intrusion sources in the vicinity of underground pipelines. Traditional techniques for detecting and pinpointing pipeline intrusions primarily rely on a single sensor monitoring point, which is susceptible to inherent errors and constraints. In contrast, the MSJLM proposed in this study leverages data from multiple samples, integrating diverse data sources through correlation analyses to elevate precision and reliability. The utilization of the CC-CNN framework for processing aggregated data has proven highly successful in extracting spatial features and identifying patterns. Furthermore, the effectiveness of this method is corroborated through validation via a pile-driving model test.