پیوندهای مفید
آمار
| تعداد نشریات | 8 |
| تعداد شمارهها | 435 |
| تعداد مقالات | 5,639 |
| تعداد مشاهده مقاله | 7,467,704 |
| تعداد دریافت فایل اصل مقاله | 6,238,210 |
Reservoir Inflow prediction Using Machine Learning Model | ||
| AUT Journal of Civil Engineering | ||
| مقالات آماده انتشار، پذیرفته شده، انتشار آنلاین از تاریخ 26 آبان 1404 اصل مقاله (796.46 K) | ||
| نوع مقاله: Research Article | ||
| شناسه دیجیتال (DOI): 10.22060/ajce.2025.23240.5866 | ||
| نویسندگان | ||
| elham Ghanbari-Adivi* 1؛ hamid hosseinkhani2 | ||
| 1Department of water engineering, Faculty of agriculture, shahrekord university, Shahrekord, Iran | ||
| 2Department of Water Science Engineering, Shahrekord University, Shahrekord, Iran | ||
| چکیده | ||
| Reservoir inflow prediction is critical for effective water management. By accurately forecasting these inflows, reservoir operators can make well-informed choices regarding water releases, which can influence both the availability of water downstream and the potential for flooding. This research introduces a novel predictive model called the Clockwork Recurrent Neural Network (CWRNN)-M5T, specifically designed to forecast monthly reservoir inflow. By synthesizing these two models, this study proposes a groundbreaking method that significantly improves prediction accuracy and provides critical insights for effective water resource management. The CWRNN-M5T model can predict inflow for one, two, and three months ahead. This study showcases the model's effectiveness, contributing to advancements in engineering informatics for water resource management and optimal dam operations. It also explores how the model's performance changes with longer prediction horizons, emphasizing its limitations and potential real-world applications. The models utilized the lagged reservoir inflow values as inputs. For one month predictions, the CWRNN model yielded the best results. However, the CWRNN-M5T model surpassed all others, achieving a Nash Sutcliffe efficiency (NSE) of 0.98, compared to 0.94 for the CWRNN model. Additionally, the CWRNN-M5T model recorded the lowest mean absolute error (MAE) at 0.123, while the CWRNN model had an MAE of 0.210. For two months predictions, the CWRNN-M5T model achieved the lowest root mean square error (RMSE) of 0.254. Overall, the CWRNN-M5T model has proven to be a highly effective tool for predicting reservoir inflow. | ||
| کلیدواژهها | ||
| Deep lerning models؛ Water resource management؛ hydrological prediction؛ hybrid model | ||
|
آمار تعداد مشاهده مقاله: 45 تعداد دریافت فایل اصل مقاله: 27 |
||