【打印本页】   【下载PDF全文】   查看/发表评论  下载PDF阅读器  关闭
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 1088次   下载 523 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于广义回归神经网络异质复合墙体日光温室温度场的预测
尹庆珍1, 张天策2, 郄丽娟1, 韩建会1
1.河北省农林科学院 经济作物研究所, 石家庄 050051;2.华北电力大学 电气与电子工程学院, 北京 102206
摘要:
为深入分析新型异质复合墙体日光温室的保温特性与应用前景,利用广义回归神经网络算法训练样本数据,通过三次样条插值法对训练结果拟合,建立冬季温室温度场预测模型。提出确定最优光滑因子的分组数目的留一优化法。选取河北省农科院经作所设计建造的新型异质复合墙体日光温室的2017年数据进行试验验证。结果表明:该模型预测效果良好,分组数目约为样本数目的1/16时训练效果最佳,预测温度与实际温度平均误差0.276 5℃,相关系数大于0.99,具有较好的精度与稳定性。本模型预测温室温度场效果良好,可用于预测冬季温室最低温度确定作物最优定植时间。
关键词:  日光温室  广义回归神经网络  留一优化法  温度场预测
DOI:10.11841/j.issn.1007-4333.2019.06.16
分类号:
基金项目:河北省第三批“巨人计划”蔬菜科研创新团队项目资助;河北省财政专项(F16R01)
Temperature field prediction and application of heterogeneous composite wall in solar greenhouse based on general regression neural network
YIN Qingzhen1, ZHANG Tiance2, QIE Lijuan1, HAN Jianhui1
1.Institute of Cash Crops, Hebei Academy of Agriculture and Forestry Sciences, Shijiazhuang 050051, China;2.School of Electrical and Electronic Engineering, North China Electric Power University, Beijing 102206, China
Abstract:
In order to thoroughly analyze the thermal insulation characteristics and application prospects of the new heterogeneous composite wall in solar greenhouse,the general regression neural network algorithm was used to process the sample data,and the results were fitted by cubic spline interpolation method.A prediction model for the greenhouse temperature field in winter was established and a retention-optimal method for determining the number of packets of the optimal smoothing factor is proposed.The data of a new heterogeneous composite wall solar greenhouse designed and constructed by Hebei Academy of Agricultural Sciences in 2017 were selected for experimental verification.The results showed that the model had good prediction effect,and the training was optimal when the number of packets is about 1/16 of the number of samples.The average error between the predicted temperature and the actual temperature was 0.276 5℃,and the correlation coefficient was greater than 0.99 indicating better accuracy and stability.In conclusion,the model can effectively predict the greenhouse temperature field and it can be used to predict the minimum temperature of greenhouse in winter for determining the optimal planting time of crops.
Key words:  solar greenhouse  general regression neural network  leave-one-out method  temperature field prediction
引用本文: