基于误差修正模型的日光温室温度预测模型  

Temperature Prediction Model of Solar GreenhouseBased on Error Correction Model

作　　者：龚健林 何斌[1,2] 张亦博 付国 Gong Jianlin;He Bin;Zhang Yibo;Fu Guo(College of Water Resources and Architectural Engineering,Northwest A&F University,Yangling 712100,China;Key Laboratory of Agricultural Soil and Water Engineering in the Ministry of Education,Northwest A&F University,Yangling 712100,China)

机构地区：[1]西北农林科技大学水利与建筑工程学院,陕西杨凌712100 [2]西北农林科技大学旱区农业水土工程教育部重点实验室,陕西杨凌712100

出　　处：《农机化研究》2023年第10期47-54,共8页Journal of Agricultural Mechanization Research

基　　金：陕西省科技创新引导专项(2021QFY08-01)。

摘　　要：日光温室作为一个复杂的非线性耦合系统,室内温度受室内外众多环境因素的影响,且大多数因素的监测数据都是非平稳时间序列。因此,在利用这些监测数据进行多元线性回归分析时,所建立的预测模型可能会出现伪回归现象导致模型预测精度较低。为此,提出了基于时间序列协整理论建立温度误差修正模型的方法。利用土后墙日光温室2021年1月收集到室内外气象数据,分别建立了典型晴天和阴天条件下的日光温室白天(9:00-18:00)逐时温度误差修正预报模型,并通过模拟回代对模型进行验证。结果显示:阴天和晴天条件下,误差修正模型的预报值与实际值相比,均方根误差分别为0.4234℃和1.5937℃,均低于传统多元线性回归模型的0.5750℃和1.9522℃,有效地提高了预测精度。Solar greenhouse is a complex nonlinear coupling system.The temperature inside the greenhouse is affected by many environmental factors inside and outside the greenhouse,and some of the factors are difficult to accurately measure and monitor.Therefore,it is extremely difficult to accurately predict the indoor temperature of the solar greenhouse.However,the existing prediction methods such as time series analysis,neural network and multiple regression analysis,which are only based on observation data,have low prediction accuracy,especially the multiple linear regression analysis method.Because these prediction method and the stability of the observed data and the accuracy is closely related,but based on the monitoring data are mostly stationary time series data,therefore,in the direct use of these data when multiple linear regression analysis,the prediction model is established and may appear false phenomenon caused by regression model prediction accuracy is low.In view of this phenomenon,this paper proposes a method to establish an indoor temperature error correction model based on the time series co-integration theory.Firstly,a long-term equilibrium model between indoor temperature,outdoor temperature,outdoor humidity,solar radiation,solar altitude Angle,wind speed,and the previous day's minimum and maximum outdoor temperature is established based on the observation data.Then,according to the co-integration theory,the co-integration relationship between the model residual and the sequence of various variables is tested,and the short-term error correction model between the indoor temperature and various variables is established.The combination of the two can more accurately achieve the purpose of predicting the indoor temperature of the solar greenhouse according to the outdoor environmental data.Based on the indoor and outdoor meteorological data collected from the solar greenhouse in Northwest A&F University of Yangling in January 2021,the hourly temperature error correction forecast models of the solar greenho

关 键 词：日光温室 温度预测 误差修正模型 线性回归 协整理论 伪回归 

分 类 号：S625.1[农业科学—园艺学]