Embedding a Long Short-Term Memory Network in a Constraint Programming Framework for Tomato Greenhouse Optimisation

Increasing global food demand, accompanied by the limited number of expert growers, brings need for more sustainable and efficient horticulture. The controlled environment greenhouses enable data collection precise control. For optimally controlling greenhouse climate, a grower not only looks at crop production, but rather aims maximising profit. However this is complex, long term optimisation task. In paper, Constraint Programming (CP) applied to task optimal economic control, leveraging learned climate model through CP embedding. collaboration with an industrial partner, we demonstrate how LSTM model, embed into framework, optimise expected profit grower. This data-to-decision pipeline being integrated decision support system multiple in Netherlands.