摘要： 为实现对烟田烟蚜发生程度的预测预报,以12年的历史资料为基础数据,采用BP神经网络方法建立了烟蚜发生程度的预测模型。该模型对待测样本的预测准确度为99.43%,回测准确度为87.36%。所建立的预测模型可提前1个多月对烟蚜发生程度进行预测,为中期预测模型,其预测结果可为烟田蚜虫综合治理提供依据。

关键词: 烟蚜, 发生程度, BP神经网络, 预测模型

Abstract: Basing on the data of 12 years, the method of BP neural network was applied to establish the predictionmodel of Myzus persicae occurrence for its prediction and forecast in tobacco field.The model exhibited prediction accura??cy of 99.43%, and back prediction accuracy of 87.36%, respectively.The prediction model could predict the occcurencedegree of Myzus persicae more than one month in advance, so the middle??term prediction model could be applied in IPMof Myzus persicae in tobacco field.

Key words: Myzus persicae, Occurrence degree, BP neural network, Prediction model

中图分类号: 

• S435.72