Smart frost control in greenhouses by neural networks models

Thermal comfort in greenhouses is a key fact to enhance productivity, due to the excess demand of energy for heating, ventilation and agroclimatic conditioning. Frost, in particular, represents a serious technological challenge if the crop sustainability is to be ensured. A Multi-Layer Perceptron artificial neural network, trained by a Levenberg-Marquardt backpropagation algorithm was designed and implemented for the smart frost control in greenhouses in the central region of Mexico, with the outside air temperature, outside air relative humidity, wind speed, global solar radiation flux, and inside air relative humidity as the input variables. The results showed a 95% confidence temperature prediction, with a coefficient of determination of 0.9549 and 0.9590, for summer and winter, respectively.