湖北农业科学 ›› 2022, Vol. 61 ›› Issue (19): 92-96.doi: 10.14088/j.cnki.issn0439-8114.2022.19.018

• 园艺·特产 • 上一篇    下一篇

基于多元回归的烟草长势影响因子分析

农英雄, 陈智斌, 黄聪, 潘剑, 梁冬, 韦屹, 李喆, 周肇峰, 陆瑛   

  1. 广西中烟工业有限责任公司信息中心,南宁 530001
  • 收稿日期:2021-09-26 出版日期:2022-10-10 发布日期:2022-11-04
  • 通讯作者: 陆 瑛(1985-),广西南宁人,工程师,硕士,主要从事企业信息化管理、大数据研究工作,(电话)18677195300(电子信箱)03429@gxzy.cn。
  • 作者简介:农英雄(1971-),男,广西天等人,高级工程师,硕士,主要从事企业信息化管理、大数据研究等工作,(电话)0771-3998605(电子信箱)00083@gxzy.cn。
  • 基金资助:
    广西科技计划项目(桂科计字〔2016〕380号); 广西中烟工业有限责任公司科技计划项目(GXZYCX2019E007)

Analysis of influencing factors of tobacco growth based on multiple regression

NONG Ying-xiong, CHEN Zhi-bin, HUANG Cong, PAN Jian, LIANG Dong, WEI Yi, LI Zhe, ZHOU Zhao-feng, LU Ying   

  1. Information Center of China Tobacco Guangxi Industrial Co., Ltd., Nanning 530001, China
  • Received:2021-09-26 Online:2022-10-10 Published:2022-11-04

摘要: 使用GreenSeeker光谱仪获取烟草的NDVI值,并采集土壤温度、土壤湿度、大气温度、风速、室外湿度和太阳辐射量等烟草生长环境影响因子值,建立偏最小二乘法回归、逐步回归、岭回归模型来探究环境影响因子对烟草NDVI值的影响,从而找到一种预测烟草NDVI值的优良方法,以便能够更加快速、准确地了解烟草的长势。试验结果表明,土壤温度、大气温度和室外湿度对烟草长势NDVI值的影响起决定性作用。根据上述回归模型的决定系数R2判断得到结论:偏最小二乘法预测精度最高,岭回归次之,逐步回归预测精度相对较差。

关键词: 烟草长势, 生长环境影响因子, NDVI, 多元回归

Abstract: The NDVI value of tobacco was obtained by GreenSeeker spectrometer, and the values of tobacco growth environmental impact factors such as soil temperature, soil humidity, atmospheric temperature, wind speed, outdoor humidity and solar radiation were collected. Partial least square regression, stepwise regression and ridge regression models were established to explore the impact of environmental impact factors on the NDVI value of tobacco growth. The purpose was to find a good method to predict the NDVI value of tobacco, so as to understand the growth trend of tobacco more quickly and accurately. The results showed that soil temperature, atmospheric temperature and outdoor humidity played decisive roles in the NDVI value of tobacco growth. According to the determination coefficient R2 of the above regression models, it is concluded that the prediction accuracy of the partial least squares method is the highest, followed by ridge regression, and the prediction accuracy of stepwise regression is relatively poor.

Key words: tobacco growth, growth environment influencing factors, NDVI, multiple regression analysis

中图分类号: