A Hybrid Model Based on LSTM for Water Prediction Algorithm

Focusing on the challenge of predicting water levels under highly nonlinear and non-smooth characteristics, we propose a data reconstruction blend prediction model based on Long Short-Term memory (LSTM) networks. The model utilizes a completely ensemble empirical model decomposition with adaptive noise to decompose the time series, and then applies the Temporal Convolutional Network (TCN) and LSTM structures as water prediction models. Through analysis of the predicted results from a pumping plant’s pumping port data, we demonstrate the superiority of our model over traditional approaches.