Multivariable Greenhouse Control Using the Filtered Smith Predictor

The increasing demand for high-efficiency greenhouse control systems motivates the interest in ensuring optimal growth climate conditions into the system. The greenhouse climate control is complex because of the strong coupling between the two main controlled variables (temperature and humidity), the time delays present within the control loop and the high nonlinear interaction between the physical and the biological subsystems. In this context, the idea of this work is to improve the robust behavior of the Filtered Smith Predictor (FSP) in an equivalent greenhouse climate model with multiple time delays as a function of the uncertainty degree of the process model to be controlled. An automated tuning technique is proposed for the robustness filter of MIMO-FSP through online estimation of the modeling error. Simulation results show that the proposed technique is able to improve disturbance rejection dynamics while assuring robust stability.