基于MATLAB 神经网络的温室草皮腾发量预测研究.pdfVIP

安徽农业科学,Journal of Anhui Agri .Sci .2008 ,36( 16) :6609 - 6610 ,6632 责任编辑 孙 红忠 责任校对 马君叶 基于 MATLAB 神经网络的温室草皮腾发量预测研究 李雪, 毛罕平, 左志宇, 张晓东, 付为国( 江苏大学江苏省现代农业装备与技术重点实验室, 江苏镇江212013) 摘要 [ 目的] 明确基于MATLAB 的BP 神经网络预测温室草皮腾发量的可行性。[ 方 法] 在9 月温室实 测气象资料的基础上, 对 温室内 的平均气温、相对湿度、光照强度和草皮 日腾发量( ET) 进行回归分析, 建立 了BP 网络ET 预报模型( BP- ET) 。[ 结果] 气温、光照强度与草 皮腾发量呈显著正相关( P 0 .05) , 相对湿度与草皮腾发量呈显著 负相关( P 0 .05) 。BP 神经 网络模型具有极 高的拟合精度,9 月资料 检验预报模型的平均相对误差为5 .58 % , 模拟与检验均有很 高的拟合精度 。BP 网络可 以用于草皮 日腾发量 的预测, 是 对传统草皮 日腾 发量计算的补充。[ 结论] 该研 究为气象数据缺测条件下温室草皮 日腾发量的估算提供 了新思路 。 关键词 温室; 作物蒸腾量; 人工神经 网络; 模型 中图分类号 S126 文献标识码 A 文章编号 0517 - 6611(2008) 16 - 06609 - 02 Study on the Prediction of Turf Evapotranspirationin the Greenhouse by Using the Neural Network Based on MATLAB LI Xue et al ( Jiangsu Provincial Key Laboratory of Modern Agricultural Equipment and Technology , Jiangsu University , Zhenj iang , Jiangsu 212013) Abstract [ Obj ective] The research ai med to confir mthe feasibility of predicting the turf evapotranspiration by usi ng BP neural network based on MAT- LAB . [ Method] On basis of the measured meteorological datai n the greenhouse in September , the regression analysis was made on the average tempera- ture , the relative humidity , light i ntensity in the greenhouse and turf evaportranspiration ( ET) to establish ET prediction model of BP network( BP- ET) . [ Result] Temperature and light i ntensity showed a significantly positive correlation with turf evaportranspiration ( P 0 .05) , while relative humidity showed a significantly negative correlation with turf evaportranspiration ( P 0 .05) . BP neural network model had an extremely high fitti ng precision . Through testing with the data in Septe mber , the average relative error of the prediction model was 5 .58 % and both sti mulati ng and testing