基于BP神经网络的病虫害预测研究  

Pest Control Prediction Based on BP Neural Network

作　　者：章银娥 张红霞[2] 黄传龙 罗晖 ZHANG Yine 1,2, ZHANG Hongxia2, HUANG Chuanlong3, LUO Hui2(1. Key Laboratory of Jiangxi Province for NumericalSimulation and Emulation Techniques ;2. Mathematics and Computer Science of GanNan Normal University;3. Ganzhou Fruit Industry Bureau, Ganzhou 341000, Chin)

机构地区：[1]江西省数值模拟与仿真技术重点实验室 [2]赣南师范大学数学与计算机科学学院 [3]赣州市果业局,江西赣州341000

出　　处：《赣南师范大学学报》2018年第3期85-89,共5页Journal of Gannan Normal University

基　　金：江西省数值模拟与仿真技术重点实验室课题;江西省科技支撑计划项目(20151BBF60071;20171BBE50065);江西省人文高校项目(JC161024)

摘　　要：建立基于BP神经网络病虫害预测预报模型,对提前采取防病防虫措施、减少农作物病虫害损失、提高农作物产量与质量具有重要意义.本研究2008-2010年的数据进行网络训练,对2011年病虫害发生程度进行回测,建立以Sigmoid函数为传递函数的BP神经网络模型.BP神经网络模型对训练样本的拟合精度有4个月超过80%,误差平方和sse为0.0180,方差mse为9.998e-004,BP神经网络模型能较好地描述病虫害的发生程度.This study aims to build up a pest control predicting model based on BP neural network, so as to prevent pest diseases, reduce crop damages by pests and disease and improve the quantity and quality of crop yields. The study has conducted training with data from 2008 to 2010 and retested the occurrence degree of crop based on the data of 2011. Finally, the study has built up a BP neural network model which takes Sigmoid function as transfer function. The fitting accuracy of BP models is above 80% in at least 4 moths, MSE values are 9. 998e-004 and SSE value is 0. 0180. , which demonstrates that the BP neural network model can well describe the degree of occurrence of plant diseases and insect pests.

关 键 词：病虫害 模型预测 BP神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]