A novel model for runoff prediction based on the ICEEMDAN-NGO-LSTM coupling

Abstract

The prediction of runoff trends has always been an essential topic in the eld of hydrological forecasting, accurate and reliable prediction models are of great signi cance to the rational use of water resources. Considering the relatively-low accuracy and poor solving ability of present models for runoff prediction, a new coupled model based on the Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN) and Northern Goshawk Optimization (NGO) with Long Short-Term Memory (LSTM) model is proposed in the paper, for runoff prediction. The present model is applied to predict the monthly runoff in the middle reaches of the Huaihe River. The results show that the ICEEMDAN-NGO-LSTM model not only increases the t by 3.09%, but also reduces the average relative error, average absolute value error, and root mean square error by 54.43, 73.11%, and 53.95%, respectively, compared with the traditional LSTM under the same data conditions. In addition, the t of the ICEEMDAN-NGO-LSTM model was improved by 0.69% and 0.56% compared with the LSTM models optimized based on Whale Optimization Algorithm (WOA) and Imperialist Competitive Algorithm (ICA), respectively, and the average relative error, average absolute value error, and root mean square error were also signi cantly lower. This indicates that the coupled ICEEMDAN-NGO-LSTM model has better prediction performance, and the prediction results provide a new method for short-term runoff forecasting.

Runoff Prediction in Different Forecast Periods via a Hybrid Machine Learning Model for Ganjiang River Basin, China

“…On the other hand, with the rapid development of decomposition methods, new methods such as the wavelet packet decomposition (WPD), complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and singular spectrum analysis (SSA) demonstrate some advantages in medium-and long-term term runoff forecasting [56,57]. However, it cannot simply be concluded that new methods will always produce the best simulations under all conditions.…”