Model predictive control of greenhouse climatic processes using on-line linearisation

Some of the objectives in greenhouse horticultural production are to increase the productivity and the quality of products and, at the same time, to reduce pollution and energy consumption. To achieve these objectives, an appropriate climate control system must be implemented. Taking into account all the process characteristics and, especially, the type of the disturbances which affect the controlled and state variables, a controller based on some optimisation criterion subject to hard input constraints appears as a convenient choice. In this paper, a Model Predictive Controller with on-line linearised models has been chosen. This nonlinear control law, which provides with a trade-off solution between performance and computational burden, is able to satisfy the specifications despite the inherent non-linearities of the system and the influence of disturbances. The application of this control strategy to commercial greenhouses in a near future is encouraged, since the simulation results show that the MPC approach clearly overcomes the performance achieved with a classical PID scheme.