Model based predictive control of greenhouse air temperature and relative humidity

This paper presents some of the work on greenhouse environmental control that has been carried out at the University of Algarve in the south of Portugal. It summarises the modelling framework and results about the models that were identified for model predictive control. Radial basis function neural networks are used as non-linear models whose parameters are determined using the Levenberg-Marquardt optimisation method, and whose structure is selected by means of multi-objective genetic algorithms. The application of the Branch-and-Bound search algorithm to discrete model-based predictive control of greenhouses is also discussed. The temperature control strategy is a mixture of temperature integration and difference between day and night temperatures. Methods were proposed to reduce the computational demand of the Branch-and-Bound algorithm and to on-line adapt the cost function coefficients, in order to increase energy savings without significantly affecting the control accuracy, by exploiting the predicted behaviour of the external climate. The methods are briefly described and a subset of simulation results are presented