Investigation and Design of Drainage Network Renovation Engineering in Residential Areas Based on BP Neural Network
Abstract
Nowadays, residential district drainage systems play a significant role in designing and investigating renovation projects to prevent water contamination and stormwater runoff. After investigating the challenges associated with existing urban drainage pipeline systems, including runoff and its impact on society and the environment, an efficient drainage network renovation is proposed, which is essential for residents’ health and quality of life. A deep learning (DL) technique called backpropagation neural network for residential drainage renovation (BPN2-RDR) is employed in district drainage network projects. Firstly, existing residential drainage infrastructures and the associated challenges are investigated. Secondly, optimized design parameters, including pipe flow, channel capacity, and slopes for controlling runoff, are adopted for the renovation strategies. The training of the neural network model with pertinent data obtained from Waipa district council, including 766 records and 36 attributes, enables the discovery of various design patterns and the identification of the relationships of the parameters within the drainage network system. The design phase utilizes the trained neural network to predict potential issues and optimize the drainage system for enhanced performance. For performance evaluation, the proposed method is analyzed using metrics such as peak flow reduction rate of the drainage system, accuracy, precision, recall, and Root Mean Square Error (RMSE). The result findings confirm the superiority of the proposed algorithm when applied to the district residential drainage network renovation
projects, thereby enhancing the residents’ quality of life.
Keywords: Backpropagation, deep learning, neural network, design parameters, residential drainage network, renovation project, stormwater.
Cite As
B. Fang, R. Hu, "Investigation and Design of Drainage Network Renovation Engineering in Residential Areas Based on BP Neural Network", Engineering Intelligent Systems, vol. 32 no. 5, pp. 521-532, 2024.