| 摘要: |
| 为建立日光温室中短期气温预报模型,以2个冬季生产季的日光温室实时气温观测资料为基础,利用BP神经网络建模和曲线拟合的方法,对日光温室1~7 d气温预报模型进行了研究。结果表明:1)以室外气温为输入要素的温室气温预报模型,最高气温预报值与观测值的符合度指数(D)为0.68~0.93,均方根误差(RMSE)为3.1~6.3 ℃;2)最低气温预报值与观测值的符合度指数(D)为0.81~0.95,均方根误差(RMSE)1.5~2.2 ℃;3)日光温室内最低气温预报绝对误差小于2 ℃的预报准确率Rate(≤2 ℃)为78%~95%;4)逐时气温预报模型预报值与实测值的符合度指数(D)为0.95~0.99,均方根误差(RMSE)为1.0~2.8 ℃,逐时气温预报模型预测准确率较高。结合目前气象台站"周预报"结果,模型可较准确地预报温室内1~7 d最低气温,并模拟日光温室内气温的逐时变化,可为冬季日光温室低温灾害预警及室内气温调控提供有益参考。 |
| 关键词: 日光温室 BP神经网络 气温 预报模型 |
| DOI：10.11841/j.issn.1007-4333.2015.01.025 |
| 投稿时间：2014-02-16 |
| 基金项目:科技部公益性行业专项(GYHY201006028,GYHY201306039); 天津市气象局科研课题(201310) |
|
| An air temperature predict model based on BP neural networks for solar greenhouse in North China |
|
LIU Shu-mei, XUE Qing-yu, LI Zhen-fa, LI Chun, GONG Zhi-hong, LI Ning
|
| (Tianjin Climate Center, Tianjin 300074, China) |
| Abstract: |
| In order to establish the greenhouse temperature prediction model for a week,a microclimate observing experiment was conducted in Xiqing,Dagang Baodi,Jinghai in the winter of 2011 and 2012.A BP neural network simulation model was established with the observational data in collected in winter of 2011,and validated the model with data from Dec.2012 to Jan.2013.The results showed that the agreement index of prediction model for the high temperature is fluctuated from 0.68 to 0.93,and the root mean square error (RMSE) is changed from 3.1 to 6.3 ℃;but that the agreement index of the prediction model for low temperature is 0.81-0.95,the root mean square error (RMSE) is from 1.5 t-2.2 ℃.The accurate rate with an error less than 6 ℃ is 60%-82% in high temperature forecasting,but 78%-95% with an error less than 2 ℃ for low air temperature.The agreement index of the real time air temperature prediction model is between 0.95 and 0.99,and the root mean square error (RMSE) is from 1.0 to 2.8 ℃.The forecast model might be applied in forecasting cryogenic disaster for 1-7 days,which combined with the weekly outsicde temperature forecasting values. |
| Key words: solar greenhouse BP neural networks air temperature prediction model |
