Flood control optimal operation of cascade reservoirs coupled with intelligent river flow algorithms

In the flood control operation of cascade reservoirs, accurate inflow calculation and efficient model solving are critical to enhancing flood control performance.Given that upstream river flow routing exerts a significant influence on the inflow process, this study addresses the flood control optimization problem of cascade reservoirs under complex hydraulic connections by proposing a flood control operation method that incorporates intelligent river flow routing.Firstly, an intelligent river flow routing model based on the Stacking ensemble learning framework was developed by integrating the multimodal advantages of BP neural networks and LSTM, which can achieve collaborative optimization of prediction results from heterogeneous models through multivariate nonlinear regression.Secondly, the Improved Grey Wolf Optimizer (IGWO) was employed for the optimal solution of the flood control operation model.By introducing Sine chaotic mapping, a nonlinear convergence factor, and a dynamic search space strategy, the efficiency of model solving was enhanced.Taking the Xiluodu—Xiangjiaba cascade reservoirs in the lower reaches of Jinsha River and the Three Gorges Reservoir as study cases, the impact of interval inflows on the flood control operation of cascade reservoirs was analyzed.The results demonstrate that the intelligent routing model outperforms traditional methods in low/high flow intervals, flood peak prediction, and trend fitting.The flood control operation scheme derived from the coupled model reduces the maximum flood regulating water level of the Three Gorges Reservoir by 0.60 m, decreases the maximum discharge by 3%, and delays the peak occurrence time by 12 hours.These findings provide valuable insights for the flood control optimal operation of reservoir groups.