A Neural Network Model to Predict Temperature and Relative Humidity in a Greenhouse

This research was developed in a interconnected polyethylene greenhouse with tomato, located in Chapingo State of Mexico, in which temperature and relative humidity were measured every five minute from January 23 to February 02 2007. During this period temperature variations were between 5 degrees C and 40 degrees C and RH oscillated between 22 -98% both of which are not within the optimal ranges, therefore a good environmental control tool is necessary to keep these variables inside of the optimal levels. Artificial Neural Network (ANN) technology was applied to predict temperature and relative humidity inside of the greenhouse because it often offers a superior alternative to traditional physical-based models, and excels at uncovering patterns or relationships in data. It is also a powerful non-linear estimator which is recommended when the functional form between input and output is unknown or it is not well understood but believed to be nonlinear. Three neural network models were implemented in the Matlab Neural Networks Toolbox, all of them with the same inputs: outside temperature (degrees C), relative humidity solar radiation (Wm-2) and wind speed (ms(-1)). The first model considered only temperature as an output with a mean square error (MSE) of 3.03 between real and predicted temperature values. The second model took into account relative humidity as an output with a MSE of 22.41 between real and predicted relative humidity values. Finally the third model considered two outputs temperature and relative humidity at the same time and produces an MSE equal to 3.39 for temperature and MSE equal to 29.23 for relative humidity. This feasibility study demonstrates that ANN technology has the potential to serve as a highly accurate forecasting tool. Moreover, ANN technology can continuously be updated, as new data become available, increasing its forecasting ability.