Abstract.
Predicting the load power of a base solar power plant is critical to improving the profit from available solar energy. This research develops a deep learning method to predict the load power utilizing the long short-term memory model. The proposed method could increase the accuracy in hourly predicting the full load power of commercial buildings integrated with a solar power plant. The load operation is generally influenced by the weather parameters such as temperature, dew point, humidity, pressure, wind direction, and wind speed. These parameters affect the load power, which is considered the target factor. Moreover, the advanced meter infrastructure also collects the internal parameters of the solar power microgrids over one year, including the battery-discharged power. These additional parameters were also examined as input parameters in the proposed method. The Long short-term memory (LSTM) algorithm is sought for predicting the full load power of the building. The performance and accuracy of the proposed LSTM are compared and evaluated with other deep learning methodologies with MSE and MAE benchmarks. The experiment results proved that the LSTM scores a higher performance with the maximum improvements of 39.64% MSE, 27.05% MAE in the training data, and 12.58% MSE, 19.94% MAE in the validating operation, respectively.
Thông tin:
| Thuộc danh mục | |
| Hội thảo quốc tế | International Conference on Sustainable Energy Technologies (ICSET2023) |
| Tổ chức | Industrial University of Ho Chi Minh City |
| Trang | 266 |
| ISBN | 978-604-920-208-7 |
| Ngày tổ chức hội thảo | 11/2023 |
| Ngày xuất bản | 11/2023 |
| Link | https://drive.google.com/drive/folders/1v806uXc_xa3_sBqvAazHURcGnnmiYvEE |