| 户雪敏1, 林汉龙2, 陈观浩3, 王春生1, 陈利军1, 史洪中1*.基于全子集回归和BP神经网络的水稻纹枯病预测模型研究[J].植物保护,2026,(1):113-120. |
| 基于全子集回归和BP神经网络的水稻纹枯病预测模型研究 |
| Prediction of rice sheath blight based on full subset regression and BP neural network |
| 投稿时间：2024-12-19 修订日期：2025-03-30 |
| DOI：10.16688/j.zwbh.2024666 |
| 中文关键词: 水稻纹枯病 发生程度 全子集回归 BP神经网络算法 |
| 英文关键词:rice sheath blight disease severity full subset regression BP neural network algorithm |
| 基金项目:信阳市科技攻关项目(20250070); 广东省科技计划（2013B020416002） |
|
| 摘要点击次数: 28 |
| 全文下载次数: 31 |
| 中文摘要: |
| 水稻纹枯病是水稻的三大病害之一, 严重影响水稻的产量和品质。本研究利用1986年－2014年广东化州地区空气温度、相对湿度、降雨量和日照时数等数据, 通过相关性分析筛选到了影响水稻纹枯病流行的9个关键因子, 并采用全子集回归和BP神经网络算法对化州地区水稻纹枯病发生程度进行预测。结果表明, 全子集回归模型1和模型2对1986年－2014年水稻纹枯病发生程度的回测准确度分别为95.6%和95.2%, 对2015年－2018年的平均预测准确度分别为90%和92.5%; BP神经网络模型1和模型2的回测准确度分别为94.5%和95.2%, 平均预测准确度均为82.5%。由上可知, 全子集回归模型的预测准确度高于BP神经网络模型, 可应用于水稻纹枯病的预测预报。 |
| 英文摘要: |
| Rice sheath blight, one of the three major diseases affecting rice, causes serious yield and quality losses. In this study, meteorological data including air temperature, relative humidity, rainfall, and sunshine hours from Huazhou, Guangdong province, collected from 1986 to 2014, were analyzed to identify nine key climatic factors influencing the epidemic dynamics of rice sheath blight through correlation analysis. Based on these factors, prediction models for disease severity were developed using full subset regression and BP (back-propagation) neural network algorithms. The results demonstrated that full subset regression models 1 and 2 achieved back-testing accuracies of 95.6% and 95.2%, respectively, for the 1986-2014 period, and average prediction accuracies of 90% and 92.5% for 2015-2018. In contrast, BP neural network models 1 and 2 showed back-testing accuracies of 94.5% and 95.2%, and average prediction accuracies of 82.5%. Overall, the full subset regression models achieved higher prediction accuracy compared to the BP neural network models, indicating their potential application in the prediction and forecasting of rice sheath blight outbreaks. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
