A Novel Framework for Optimization and Evaluation of Sensors Network in Urban Drainage System

Abstract

Efficient management of urban drainage system (UDS) is crucial for understanding the operating status of UDS and facilitating urban flood early warning. Establishing an appropriate sensor network is fundamental to achieving cost-effective and sustainable management of UDS. Previous researches have predominantly focused on optimizing sensor placement but have often overlooked the evaluation of sensor network performance. To address this gap, we propose a framework that not only optimizes sensor placement using information theory but also evaluates the performance of sensor networks through matrix completion. After the method was tested in the case study, we found that the information amount provided by the selected nodes and information redundancy among nodes in UDS can both be effectively represented by the information theory approach, and then optimal sensor networks with different numbers of sensors was selected. Furthermore, the matrix completion algorithm successfully evaluated the sensor network's performance in operation status perception and flooding risk assessment. The results indicated that the operation status perception error was 33%, and the flooding risk assessment accuracy reached 76% with four sensors. Increasing the sensor count to eight reduced the error to 29% and improved accuracy to 82%.Thus, it is evident that the matrix completion algorithm is a rapid and accurate method for evaluating sensor network performance. This study provides a comprehensive framework for sensor network optimization and evaluation, which can greatly facilitate the development of urban flood risk early warning and sustainable management of UDS.