Leveraging Genetic Algorithms for Enhanced Convolutional Neural Network Performance in Detecting Pipeline Leaks: A Case Study of Ede Water Works, Nigeria

This study explores the integration of Genetic Algorithms (GA) with Convolutional Neural Networks (CNN) to enhance the accuracy and efficiency of detecting pipeline leaks within water distribution networks. Traditional leak detection methods often face challenges due to aging infrastructure and complex environments, leading to inefficient water management. The study specifically focuses on Ede Water Works in Nigeria, utilizing EPANET to simulate various leak scenarios and collect time-series data, which is then preprocessed and segmented for CNN model training. A GA is employed to optimize the CNN’s hyperparameters, resulting in an enhanced model with improved accuracy and reduced computational cost. The optimized CNN model showed significant improvements over manually tuned models, with a training accuracy of approximately 94% and a test accuracy of 91%. The integration of GA with CNN provides a robust approach for leak detection, contributing to the sustainability and efficiency of water distribution systems.