Model Predictive Control for the Scheduling of Seedings in an Adaptive Vertical Farm

A model predictive control approach is presented for the scheduling of sowings in an adaptive vertical farm, i.e., an innovative vertical greenhouse in which the spacing between shelves is automatically adapted to crop growth. First, a dynamic model describing the evolution of occupancy and shelf height is developed. The model is affected by disturbances to account for possible deviations of crop growth from the nominal pattern. Then, an optimal control problem over a given timeframe is defined to determine the best time instants to perform seedings in the various shelves with the goal of maximizing production yield. The repeated solution of the optimal control problem over a shorter, moving window over time, according to the receding horizon paradigm, allows to devise robust control strategies with respect to disturbances, even in the absence of predictions about their future realizations. Preliminary simulation results are reported for different control horizons and type of disturbances to showcase the effectiveness of the proposed approach in maximizing production yield while exploiting almost all the available vertical space.