Temperature and Humidity Dataset of an East-Facing South African Greenhouse Tunnel

The aim of this study was to model the thermal characteristics of an east-facing greenhouse tunnel in South Africa. The tunnel used has a fan and wet wall on opposite ends and is used to cool the tunnel when temperatures reach a certain threshold. In this study, the authors' research hypothesis was the development of accurate empirical and analytical thermal models that would be validated using the available measured data. Three temperature and humidity sensors were placed near the fan, the wet wall, and in the middle of the 29-meter-long tunnel. The middle sensor's temperature measurements were used to control when the fan turned on and off. When the temperature reached above 30 degrees Celcius the fan was turned on, and when the temperature reached 22 degrees Celcius the fan was turned off. This created an 8-degree Celcius hysteresis band in which the fan and wet wall was controlled. The ``Training Set.csv`` file was used to train a Support Vector Regression (SVR) model to predict and simulate temperatures an hour in advance using previous predictions. ''Test Set.csv'' was used to validate and measure the accuracy of the SVR model. Also, the analytical model's accuracy was measured using the ``Test Set.csv`` file to quantitatively compare the two models. The test set and training set encompasses 42 days of measurements, aggregated with solar radiation and ambient temperature measurements that were captured by MeteoBlue (www.meteoblue.com). As this weather data was in an hourly format and the three temperature and humidity sensor data was in 5-minute intervals, the data from MeteoBlue was linearly interpolated for improved model training. For the ``Full Data Set.csv``, the data includes 162 days of only the three sensor's temperature and humidity measurements, and the fan and wet wall state.

本研究旨在为南非朝东的隧道式温室大棚构建热特性模型。该隧道两端分别设有风机与湿帘，当内部温度达到设定阈值时，将启动系统以降低隧道内温度。本研究的研究假说为构建精准的经验热模型与解析热模型，并利用现有实测数据完成模型验证。研究人员在风机侧、湿帘侧以及29米长隧道的中部布设了3台温湿度传感器，其中中部传感器的温度监测数据用于控制风机启停：当温度高于30℃时启动风机，当温度降至22℃时关停风机，由此形成8℃的滞回控制区间，以此调控风机与湿帘系统的运行。本研究使用"Training Set.csv"训练支持向量回归（Support Vector Regression, SVR）模型，以利用前期预测数据提前1小时预测并模拟隧道内温度；使用"Test Set.csv"对该SVR模型的精度进行验证与评估。同时，本研究亦利用"Test Set.csv"测试解析模型的精度，以实现两种模型的定量对比。训练集与测试集涵盖了42天的实测数据，同时整合了由MeteoBlue（www.meteoblue.com）采集的太阳辐射与环境温度数据。由于该气象数据为小时级格式，而3台温湿度传感器的采样间隔为5分钟，因此对MeteoBlue的气象数据进行了线性插值处理，以优化模型训练流程。"Full Data Set.csv"则包含了162天的仅来自3台传感器的温湿度数据，以及风机与湿帘的运行状态数据。