Multiple neural control of a greenhouse

In this paper an ART2 classifier is used to extract local models of a database taken from a greenhouse. Once the clusters are formed, multilayer feed-forward neural networks are then trained to model each cluster (subsystem) in order to achieve a multiple neural control of the greenhouse. The considered control strategy consists of the division of the greenhouse control phase in periods where a suitable controller is selected to drive the internal climate of the greenhouse, which is modeled with an Elman neural network. The same ART2 classifier is then used as a supervisor to select the suitable neural controller corresponding to the appropriate mode.