An optimal and intelligent control strategy to ventilate a greenhouse

In this paper, it is presented an optimal and intelligent technique to ventilate a greenhouse during the day. This technique is the result of the combination of a neural network and the particle swarm optimization algorithm. First, predictions on the dynamic behavior of the system variables are computed by means of a multilayer recurrent perceptron, trained with an extended Kalman filter. Then, using these predictions and the particle swarm optimization algorithm, we calculate the time instants when the fans of the greenhouse must be turn on and off, in order to eliminate the unwanted excess of temperature and at the same time minimizing the time lapse where the fans remain turned on. The algorithm performance is shown through simulation.