Multi-Step Ahead Time-Series Forecasting of Sediment Load Using NARX Neural Networks
River sedimentation is a universal issue in a river catchment. It can affect the reservoir ability, the river flow, and dam structure including the hydropower capacity. Therefore, having multi-step ahead forecasting for the sediment load is beneficial in terms of research and applications. This stud...
Saved in:
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Conference Paper |
| Published: |
Springer Nature
2024
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | River sedimentation is a universal issue in a river catchment. It can affect the reservoir ability, the river flow, and dam structure including the hydropower capacity. Therefore, having multi-step ahead forecasting for the sediment load is beneficial in terms of research and applications. This study discusses and presents a case study in multi-step ahead forecasting for the sediment load using non-linear autoregressive with exogenous inputs (NARX) neural networks. We use sediment data that was recorded from 8 locations in the Ringlet reservoir (upstream sections) in Malaysia. The results suggest that the NARX neural networks have good capability to do multi-step ahead forecasting for sediment load in a recursive way (closed-loop mode) based on its past values and the past values of suspended solid and discharge. The model is evaluated with performance metrics yielding NSE = 0.99 (Nash�Sutcliffe efficiency coefficient) for both the training and test dataset, and RMSE (root means square error) of 0.22 and 0.25, respectively, training and test dataset. � 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG. |
|---|