Study on Winter Wheat Water Requirement Prediction based on CNN-BiLSTM-Attention Model

Abstract

Accurate prediction of winter wheat water requirements is crucial for improving water resource utilization efficiency,increasing farmers' income,and ensuring national food security.To enhance the accuracy of winter wheat water requirement prediction,a CNN-BiLSTM neural network prediction model incorporating an attention mechanism (CNN-BiLSTM-Attention) is proposed in this paper.This model utilizes a convolutional neural network to extract spatial features of meteorological factors,employs a bidirectional long short-term memory network to learn long-term temporal dependencies of water requirements,and introduces an attention mechanism to focus on key growth stages influencing water demand formation.To validate the model's performance,a study was conducted based on daily meteorological data from Zhengzhou.The results indicate that the predictions of the CNN-BiLSTM-Attention model are closer to the benchmark truth and outperform five comparison models,including BP,LSTM,and BiLSTM,across various evaluation metrics.This demonstrates higher prediction accuracy and indicates that the model has good regional applicability,providing a basis for crop water requirement forecasting and sustainable agricultural development.