基于深度学习LSTM多模型的湖南农业产量预测研究

Agricultural Yield Prediction Based on Deep Learning LSTM Multiple Models in Hunan Province

农产品产量预测是农业科学研究的一个重要领域,对于农业生产的规划、决策和管理具有重要意义。随着神经网络预测模型的快速发展,深入研究其在农业领域的预测应用成为热点。为此,提出了能够有效捕获数据长时间依赖关系的农产品产量预测LSTM模型,在此构建了单层LSTM、堆叠LSTM、双向LSTM和BP模型,对湖南农产品油料、粮食和棉花进行数据收集、特征提取、模型训练、评估和预测。实验表明:第一,4种模型都获得了较好的结果;第二,模型性能不仅仅与模型结构、参数有关,也与数据集多少直接相关;第三,构建的模型可以用于湖南的农产品产量预测应用。

Crop yield prediction is an important area of agricultural scientific research and holds significant implications for agricultural production planning,decision-making,and management.With the rapid development of neural network prediction models,exploring their predictive applications in the agricultural field has become a hot topic.To effectively capture long-term dependencies in data,this study proposes a single-layer Long Short-Term Memory(LSTM)network model for crop yield prediction.In this context,LSTM,stacked LSTM,bidirectional LSTM,and backpropagation(BP)models are constructed,and data collection,feature extraction,model training,evaluation,and prediction are conducted for oil crops,cereals,and cotton in Hunan province.The experiments demonstrate the following:Firstly,all four models achieve good results.Secondly,model performance is not only related to model structure and parameters but also directly influenced by the size of the dataset.Lastly,the constructed models can be applied to crop yield prediction in Hunan province.