| Title | AI Control Methodology Applied in Geothermal District Heating System |
|---|---|
| Authors | Ming Li, Caixia Sun |
| Year | 2026 |
| Conference | World Geothermal Congress |
| Keywords | |
| Abstract | Geothermal energy systems provide an eco-friendly, cost-effective solution for district heating, driving urban decarbonization and sustainable development. Geothermal district heating systems are complex, nonlinear systems influenced by various factors including weather conditions, system design, user behavior, building types, and thermal losses. Current operational strategies mainly rely on weather compensation adjustments, resulting in suboptimal thermal performance and low energy efficiency. This challenge necessitates smart system optimization that balances energy efficiency with real-time performance through predictive control. This study develops a Long Short-Term Memory (LSTM) neural network incorporating meteorological data, system operational parameters, and indoor temperature measurements. Time series values of these parameters serve as input features to capture system thermal lag and building thermal inertia. The modeling process consists of three key steps: 1) Training the model using meteorological data and system operation data as inputs and indoor temperature as target variable; 2) Setting the indoor temperature control interval and calculating the appropriate water supply temperature; 3) Calculating the operating status of system components including submersible pumps, heat pumps, and circulation pumps. After validation, the trained neural network model is integrated into the SCADA system, enabling automated operation of geothermal systems. |