Data-driven models often neglect the underlying physical principles, limiting generalization capabilities in water distribution systems (WDSs). This study presents a novel spatio-temporal graph physics-informed neural network (ST-GPINN) for water quality prediction in WDSs, integrating hydraulic simulations, physics-informed neural networks (PINNs), and graph neural networks (GNNs) to capture dynamics and graph-based network connectivity while approximating partial differential equations (PDEs). ST-GPINN discretizes WDSs using virtual nodes to enhance spatial granularity, employs an Encoder-Processor-Decoder architecture for predictions. Validated on Network A (a small-scale network with 9 junctions and 11 pipes) and Network B (a real large-scale WDS with 920 junctions and 1032 pipes), ST-GPINN outperforms others, achieving a MAE of 0.0073 mg/L, RMSE of 0.0121 mg/L, and R2 of 88.91% in Network A, and a MAE of 0.008 mg/L, RMSE of 0.0098 mg/L, and R² of 98.91% in Network B. Its scalability and accuracy highlight ST-GPINN’s potential for water quality predictions.
The influence of the outer reinforcement layers on the repair structure and aerodynamic performance was studied. Firstly, a continuous damage mechanics model was established, and the 3D Hashin criterion and cohesive zone material model were used to analyze the damage repair model. The failure load deviation was 5.5%. Secondly, on the basis of the transition model and SST–ω turbulence model, the aerodynamic analysis model of DU300 airfoil was established. The numerical simulation results showed that the lift coefficient and pressure distribution at the angle of attack of 10° and 15° were deviated from the experimental values by 2%. Furthermore, 27 structural repair models, nine 2D aerodynamic repair models, and a 3D full-scale blade model were designed. It was found that, when the repair length accounted for 60% of the total model length, the failure load increased by 22%, but the aerodynamic power with the repair length of 10 m was decreased by 0.137%. When the repair area was large and the repair height was from 4 mm to 6 mm, the failure load was greatly increased by about 30%, and the aerodynamic pressure distribution and static pressure field fluctuated significantly. The results show that the structural and aerodynamic characteristics were closely related to the repair parameters.
18 December 2015 in Journal of Wuhan University of Technology-Mater. Sci. Ed.