Application of artificial neural network to forecast the water level at xuan quan culvert in viet nam

Abstract

Artificial intelligence is increasingly widely used in many different areas of life, including irrigation. In recent years, Machine learning and Artificial neural networks (ANN) models are excellent tools for forecasting flow rate in the river or the water levels at irrigation culverts to help the operation of the sluice gates be appropriate and effective. In this study, a simple and efficient model based on ANN is suggested for water level prediction at Xuan Quan culvert in Vietnam for 6 hours, 12 hours, 18 hours and 24 hours of lead time. The input data needed for prediction are observed water levels in the upstream and downstream of the Xuan Quan culvert, from 2003 to 2018. Although the model does not require a lot of data, the forecasted results are highly accurate. The Nash–Sutcliffe efficiency (NSE) reaches 97%-99% for all forecasting cases. This result shows that the ANN model proposed by the research team accurately predicts water levels in real time, furthermore, this model can be applied to forecast water levels at other irrigation sluice gates.