A Waterlogging Defense Strategy With Cross-Domain Data-Driven Risk Prediction for Coupled Power and Drainage Systems

The frequent occurrence of rainstorm-induced waterlogging has brought new requirements and challenges to the safe operation of urban power systems (UPSs) and urban drainage systems (UDSs). This paper proposes a waterlogging defense strategy with cross-domain data-driven risk prediction to reduce the flooding and power outage risk of coupled UPSs and UDSs. The optimal pre-positioning locations for mobile generators and drainage vehicles are determined to improve emergency preparedness before extreme rainfall, while synergistic dispatching of UPS network reconfiguration and dispersed sluice-pump-reservoir clusters are performed for service restoration of the power supply and waterlogging prevention during rainstorms. A data-driven waterlogging risk prediction model with a cross-domain attention fusion module is developed to estimate the potential outage scale in UPSs under waterlogging threats by integrating space-air-ground multi-source data. To boost emergency decision-making efficiency for coupled UPSs and UDSs, an equivalent representation method reformulates the UDS model with nonlinear hydrodynamic partial differential equations (PDEs) into a general state distribution matrix by eliminating intermediate variables. Comparative results have validated the superiority of the proposed strategy in mitigating the load curtailment and inundation depth.