| Keywords |
Deep Neural Network(DNN); Cooling water temperature; Algorithm; Field validation; Energy Cost |
| Abstract |
This study presents the development and field application of a DNN-based control algorithm for optimizing cooling water temperature to reduce cooling energy consumption in an office building. The DNN prediction model was trained using actual operating data and validated in accordance with ASHRAE Guideline 14. An optimization algorithm was then developed to determine the cooling water temperature that minimizes the operating cost of the heat source system, and the algorithm was deployed in the building’s HVAC system.
Using 2022 data as the baseline, the algorithm was applied during the 2023 and 2024 cooling seasons, resulting in operating cost reductions of 26.4% and 19.5%, respectively. The findings, based on three years of operational data, demonstrate the effectiveness of the proposed algorithm in achieving energy cost savings under real building conditions. |