Session: 02-05-01: Data Driven Design

Paper Number: 145244

145244 - Enhancing Water Leakage Detection in Transmission Lines Through Deep Learning-Analyzed Ground Penetrating Radar Images 

The effective management and maintenance of water transmission lines are critical to ensuring the reliability and efficiency of water supply systems worldwide. Water leaks from these systems have serious negative effects on the environment and economy in addition to wasting a lot of resources. While somewhat successful, traditional techniques for finding water leaks frequently lack precision, effectiveness, and early detection capabilities. This paper addresses these issues by introducing a unique method for detecting water leaks in transmission water lines using deep learning algorithms to analyze Ground Penetrating Radar (GPR) images.

Ground Penetrating Radar (GPR) technology offers a non-invasive means of obtaining subsurface images, providing valuable data on the condition of water pipelines. However, it takes specialist knowledge and a lot of effort to interpret GPR images correctly in order to locate water leaks. We provide a framework that automatically analyzes GPR pictures, improving detection speed and accuracy by utilizing deep learning advances. Our approach entails processing and categorizing GPR pictures as suggestive of either the presence or absence of water leaks using Convolutional Neural Networks (CNNs), a family of deep learning algorithms particularly skilled at evaluating visual data.

The paper was divided into multiple stages. The first involved gathering GPR data from our transmission water pipe setup that had controlled leakage. The robustness and generalized performance of the model were then enhanced by applying data augmentation to the dataset. Subsequently, this dataset was used to train and verify a CNN model, with special focus on architecture optimization for the unique features of GPR images associated with water leakage.

Our findings show that the model outperforms conventional techniques in terms of detection skills, exhibiting a high degree of accuracy in identifying water leaks. The early discovery of leaks can be greatly aided by the system's accurate analysis, which will save water loss and the related expenses of late-stage repair. Moreover, this approach presents the possibility of ongoing surveillance and evaluation of the condition of water transmission infrastructure, hence promoting more environmentally friendly water management techniques.

The paper addresses the implications of implementing deep learning technology in the context of water infrastructure maintenance in addition to outlining our approach and results. It also discusses potential drawbacks, such as the requirement for large amounts of data for model training and issues with the interpretability of model conclusions. This work establishes a precedent for the use of cutting-edge analytical approaches in the upkeep of vital infrastructure in addition to demonstrating the potential of combining deep learning with GPR technology for the detection of water leaks.

Presenting Author: Mariam Alnaqbi Dubai Electricity and Water Authority - Research and Development

Presenting Author Biography: Mariam is a distinguished professional in the field of Artificial Intelligence and Machine Learning, having completed her Master's degree at the prestigious University of Birmingham. Her academic journey in AI and Machine Learning has endowed her with a profound understanding and expertise in these cutting-edge technologies, enabling her to contribute significantly to the advancement of these fields.

Currently serving as a Research and Development Technologist at Dubai Electricity and Water Authority (DEWA), Mariam plays a pivotal role in driving innovation and technological development within the organization. Her work at DEWA involves exploring and implementing AI and machine learning solutions to enhance operational efficiencies and sustainability, marking her as a key player in the organization's pursuit of technological excellence.

In addition to her professional achievements, Mariam has also made a substantial impact as a Coding Ambassador for the Artificial Intelligence, Digital Economy, and Remote Work Applications Office in the UAE. In this role, she is dedicated to promoting coding literacy and AI awareness, fostering a culture of innovation and technological empowerment. Her efforts in this capacity underscore her commitment to advancing the digital economy and supporting the UAE's vision of becoming a global leader in the field of AI and digital technologies.

Mariam's contributions to the field of AI and her active involvement in promoting coding education reflect her passion for technology and her dedication to empowering others through knowledge. Her work not only enhances the technological capabilities of DEWA and the UAE but also inspires future generations to explore and innovate within the realms of artificial intelligence and digital technology.

Authors:

Mariam Alnaqbi Dubai Electricity and Water Authority - Research and Development
Mahmoud Rezk Dubai Electricity and Water Authority - Research and Development
Khuloud Almaeeni Dubai Electricity and Water Authority - Research and Development
Fahed Ebisi Dubai Electricity and Water Authority - Research and Development