Hydrological Time Series Prediction Model Based on Attention-LSTM Neural Network

Constructing predictive models with neural networks has always been the focus of research in chaotic time series prediction. Since Hinton proposed the concept of deep learning in 2006, the development of neural networks is getting faster and faster, and a variety of neural networks appear. Among them, RNN neural network and LSTM neural network are applied to all fields of our daily life for the property of well-processed time series. Therefore, we use LSTM neural network to construct our model, and optimize the model by Attention mechanism to establish an Attention-LSTM hydrological time series prediction model. The experimental results show that the Attention-LSTM model has better improvement in the mean square error and absolute error of the predicted values than the common LSTM model and the traditional BP model. And due to the introduction of the Attention mechanism, it can highlight the key factors to some extent. influences.The experimental results show that the Attention-LSTM model has the advantages of high prediction accuracy and small lag error, which is helpful for the application of deep learning algorithm in hydrological time series prediction.