116 - State Estimation-Based Placement of Water Quality Sensors in Water Distribution Systems

Water distribution systems (WDSs) play a critical role in ensuring the safe and efficient delivery of clean water to homes, businesses, and public institutions. For this purpose, continuously collecting and analyzing monitoring data is crucial. This process includes monitoring hydraulic and water quality parameters in WDSs, including water flow, pressure, and chlorine concentration. However, WDSs are large and complex infrastructure networks, including varying distribution points and pipe sizes, which makes it challenging to collect data from all points of the network. One of the critical aims of WDS monitoring is to enable state estimation (SE), which involves the use of mathematical models to determine unknown variables in the system based on data collected from monitoring sensors and hydraulic/water quality modeling. Nevertheless, SE-based metrics were rarely used in finding the optimal placement of water quality sensors in WDSs. This study aims to develop a neural network (NN)-based machine learning (ML) model for water quality state estimation and sensor placement optimization within WDSs. The model will be trained using a database that includes water demand and chlorine concentrations at each sensor location. Bayesian Optimization (BO) is implemented to determine the optimal locations for sensor placement. The objective function is formulated to maximize the accuracy of the ML-based water quality state estimation at each node within the WDS.