Optimizing monitoring strategies for urban drainage systems via bilayer iterative clustering.

Online monitoring is increasingly essential for the effective management and operation of urban sewer systems, yet resource limitations necessitate careful planning of sensor deployment. This study aims to address the impact of time lags on monitoring point selection in urban drainage systems using unsupervised machine learning techniques. A novel method is introduced to determine the optimal number and placement of sensors in manholes, using cluster analysis informed by simulated time-series data. The proposed methodology involves two sequential stages: the first stage clusters time-series data based on morphology similarity using the time-lagged cross-correlation (TLCC) coefficient, which measures the temporal alignment between datasets. The second stage further refines these clusters by considering magnitude similarity, employing dynamic time warping distance to quantify shape-based similarities and improve clustering accuracy. The proposed approach allows for flexible threshold adjustments to accommodate specific engineering requirements, enabling the design of monitoring strategies tailored to a predetermined number of locations. Furthermore, the study explores the impact of rainfall intensity on sensor placement, providing actionable guidance for sewer managers to improve monitoring efficiency and address urban water management challenges.