Currently, there are greenhouse models that can be used for the management of the green- house, including climate, energy and crop models. However, the crop models are currently pa- rameterized by hand and reflect an average crop model, thus not very realistic to the actual situation in the greenhouse. In this feasibility study, we will investigate the possibility to use and develop automatic sensing systems, including 3D machine vision, in order to estimate plant parameters of every plant in the greenhouse, coupled to functional-structural plant models for accurate simulation of the plant development and function of every individual plant in the green- house. Coupling the real-time measurements and individual plant models to the greenhouse- climate and energy models, will allow a much more accurate prediction and control of the green- house.
We suggest a Digital Twin of a greenhouse, including greenhouse-climate and energy simulation, a climate-control system and crop growth and development models based on a 3D plant model, which responds to real-time measurements and management strategy (figure 1). This Digital Twin avoids laborious and possibly error-prone crop monitoring by humans. Its application in horticultural production systems is attractive for at least two reasons: (1) high-tech greenhouses offer a clean, dry and accessible environment for advanced monitoring equipment, (2) climate control is highly advanced and can accurately address the crop requirements when monitored. The status of the crop is monitored with state-of-the-art sensors in order to feed and update the models. Important crop parameters, like stem thickness, leaf area, internode length, fruit size and ripeness, are measured frequently per plant with 3D-vision systems. They are used as inputs to functional- structural plant models to generate a 3D representation of every plant in the greenhouse. These simulated plants are then used in the greenhouse climate model for model- ling of light interceptance, evaporation and photosynthesis being the source of assimilates and growth.
Our Digital Twin (DT) is hierarchical and consists of a greenhouse model including many plant models. On basis of the expected outside climate and control parameters, a prediction of the production for the coming period (days, weeks, months) can be made. These predictions can be used to improve the decision support for climate control and for the planning of green- house operations, such as yield prediction, planning of harvesting operations and planning of leaf picking. Knowing the calculated light interceptance of each leaf, an estimation of the net
contribution per leaf to the crop production can be made, which results in an optimal leaf picking strategy for the coming period.
|Effective start/end date||1/01/19 → 31/12/19|