DeepAUP: A Deep Neural Network Framework for Abnormal Underground Heat Transport Pipelines
- 28 June 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 21 (2) , 2017-2026
Abstract
Pipeline inspection methods are continuously being developed, given that the safe maintenance and operation of heat transport pipelines is one of the most important issues in Korea. However, owing to the limitations of traditional inspection methods (e.g., excavation work), new solutions for underground heat-transport pipelines are required. Therefore, we propose DeepAUP, which is a deep neural network framework for detecting abnormalities in underground heat-transport pipelines. With datasets collected from actual environments using two sensors, DeepAUP effectively identified abnormal underground heat-transport pipelines, with more than 99% accuracy. Based on the experimental findings, several practical implications, as well as notable limitations, are examined. Note to Practitioners —This study was motivated by the several issues related to checking and maintaining underground heat-transport pipelines. Although there are several cornerstones for addressing this issue, they involve notable limitations, including economic and environmental concerns. Most existing studies indicate that expensive additional facilities, such as hoop strain sensors, are required to explore underground pipeline abnormalities. Thus, this study suggests a unique approach using deep learning called DeepAUP, which is expected to significantly reduce the effort required from practitioners. In future studies, we plan to employ a more efficient framework and a real-time detection system in the urban areas of South Korea.Keywords
Funding Information
- Ministry of Science and ICT (MSIT), South Korea
- Institute of Information & Communications Technology Planning & Evaluation (IITP-2020-0-01816)
- Ministry of Land, Infrastructure and Transport (21ATOG-C161932-01)
This publication has 48 references indexed in Scilit:
- A Hybrid Intelligent Approach for Metal-Loss Defect Depth Prediction in Oil and Gas Pipelines2016
- Probabilistic multiple model neural network based leak detection system: Experimental studyJournal of Loss Prevention in the Process Industries, 2015
- Signal processing and defect analysis of pipeline inspection applying magnetic flux leakage methodsIntelligent Service Robotics, 2014
- An improved PSO-SVM model for online recognition defects in eddy current testingNondestructive Testing and Evaluation, 2013
- Oil and gas pipeline failure prediction system using long range ultrasonic transducers and Euclidean-Support Vector Machines classification approachExpert Systems with Applications, 2013
- Acoustic emission pattern recognition approach based on Hilbert–Huang transform for structural health monitoring in polymer-composite materialsApplied Acoustics, 2013
- Defect Characterization With Eddy Current Testing Using Nonlinear-Regression Feature Extraction and Artificial Neural NetworksIEEE Transactions on Instrumentation and Measurement, 2013
- Angular and axial evaluation of superficial defects on non-accessible pipes by wavelet transform and neural network-based classificationUltrasonics, 2010
- PIPELINE DEFECT PREDICTION USING SUPPORT VECTOR MACHINESApplied Artificial Intelligence, 2009
- An improved automated ultrasonic NDE system by wavelet and neuron networksUltrasonics, 2004