Development of a Dam Management Support System Combining Long-Term Inflow Forecasting and a Td3 Based Release Operation Model

Key Points

• This research aims to develop a dam management support system that utilizes AI for better flood control through inflow forecasting and release operations.
• Developed inflow forecasting and release operation models based on existing research.
• Utilized TensorFlow for model construction and parameter exploration.
• Applied the models to the target dam and validated their effectiveness using data from the October 2019 heavy rainfall.
• The inflow forecasting model achieved sufficient accuracy when compared to actual measurements.
• The release operation model consistently implemented a fixed release rate.