A hybrid-controlled approach for maintaining nocturnal greenhouse temperature: Simulation study

[Display omitted] •A greenhouse temperature hybrid model is developed based on linear reduced models.•An HMPC is designed to control greenhouse temperature using two heating systems.•Simulations results show that the HMPC framework is suitable to solve the greenhouse heating problem. Forced-air heaters and aerial pipe systems are the most common heating equipment used in greenhouses to control the nocturnal temperature in Mediterranean areas. These heating systems are often used separately and they are seldom combined in the same greenhouse for temperature control purposes. The main reasons are that the advantage of combining both heating systems has not been thoroughly analysed in literature, and that, the complexity of the problem increases from a control point of view due to the mixing of different dynamics. The combination of these two heating systems can be useful in some situations, obtaining a reasonable trade-off between thermal gain and running costs. Thus, this paper proposes to analyse the combination of these two heating systems and provide a solution to the problem of switching two different heating systems to control the nocturnal temperature in a greenhouse by using a hybrid controller. The proposed controller counteracts the switching disadvantages presented by commercial systems based on heuristic rules. To achieve this solution, the system dynamics are represented through a hybrid model, where weather variables act as logical conditions to switch between the different process dynamics. This approach allows considering the greenhouse dynamics as a hybrid system with continuous and discrete components. A Model Predictive Hybrid Controller is used to regulate the inner temperature during the night and calculates optimal control signals based on power consumption and commutation minimisation. The performance of this controller is studied, comparing its reference deviation, number of commutations, and running costs against commercial controllers. The final results show that the adequate combination of these heating systems can contribute to a much better control performance with a minor cost increment.