基于判别分析的玉米茎腐病发生程度预测模型  

Prediction model of corn stalk rot based on discriminant analysis

作　　者：张家齐[1] 张毅[2] 石洁[1] 张笑晴 郭宁[1] 张海剑[1] ZHANG Jiaqi;ZHANG Yi;SHI Jie;ZHANG Xiaoqing;GUO Ning;ZHANG Haijian(Institute of Plant Protection, Hebei Academy of Agricultural and Forestry Sciences, Integrated Pest Management Center of Hebei Province, Key Laboratory of Integrated Pest Management on Crops in Northern Region of North China, Ministry of Agriculture and Rural Affairs, Baoding 071000, China;The National Agro-Tech Extension and Service Center, Beijing 100125, China)

机构地区：[1]河北省农林科学院植物保护研究所,农业农村部华北北部作物有害生物综合治理重点实验室,河北省农业有害生物综合防治工程技术研究中心,保定071000 [2]全国农业技术推广服务中心,北京100125

出　　处：《植物保护》2020年第2期85-90,共6页Plant Protection

基　　金：国家重点研发计划(2018YFD0200602);国家现代农业产业技术体系(CARS-02)。

摘　　要：对河南省夏玉米区2003年-2018年玉米茎腐病发生、流行情况进行分析,选择制约茎腐病发生流行的主要气象因素,采用判别分析法建立茎腐病发生的两阶段预测模型。应用2003年-2014年的数据进行模型训练,利用2015年-2018年数据进行测试。结果表明,第一阶段预测模型历史回代自身验证准确率为86%,交互验证准确率为79.1%;第二阶段自身验证准确率为83.7%,交互验证准确率为76.7%。应用2015年-2018年数据进行测试,第一阶段预测准确率75%,第二阶段预测准确率85%。说明利用判别分析建立玉米茎腐病流行模型是可行的。We selected the main factors as restrictive elements for the studies of occurrence and epidemics of corn stalk rot in Henan province from 2003 to 2018.By using the method of discriminant analysis,we built up a two-stage prediction model.We set up the model based on the data of 2003-2014 and used the data of 2015-2018 to test the model.The results showed that the accurate rate of the first-stage prediction model was 86%in self-validation and 79.1%in cross-validation;the accurate rate of the second-stage prediction model was 83.7%in self-validation and 76.7%in cross-validation.Tested with the data of 2015-2018,the accurate rates of the first and second prediction models were 75%and 85%,respectively.The results also proved a feasible method for predicting corn stalk rot by using discriminant analysis.

关 键 词：玉米茎腐病 判别分析 预测模型 

分 类 号：S431.11[农业科学—农业昆虫与害虫防治]