Artificial Neural Network Model of Plant Shoot-tip Temperature in Catharanthus roseus

An artificial neural network (ANN) model that can be used to predict the shoot-tip temperature of Vinca (Catharanthus roseus) is presented. The ANN architecture consists of a threelayer network, input data (four environmental conditions of drybulb, wetbulb, glazing temperatures and shortwave radiation) and output data (shoot-tip temperature). Data for training and validation were collected every 10 seconds and 10-minute averages for at least 41 days were stored in a computer, and subsets of these data were used for training. Validation studies indicated excellent generalization over the range of obtained data. Simulation studies with the developed model were performed to evaluate the effect of environmental factors on plant shoo -tip temperature, and it became clear that drybulb had the highest contribution ratio (88%) and shortwave radiation had the lowest under our environmental conditions. The proposed model can be applicable because its inputs consist of four environmental factors that are easily and/ or commonly measured in commercial greenhouses, and may thus be a useful tool for evaluating the environmental factors that affect plant shoot-tip temperature under greenhouse conditions.