Simulation-based control of enclosed ecosystems - A case study: Determination of greenhouse heating setpoints

Lacroix, R. and Kok, R. 1999. Simulation-based control of enclosed ecosystems - A case study: Determination of greenhouse heating setpoints. Can. Agric. Eng. 41:175-183. The overall objective of this study was to investigate the use ofa simulation-based approach for the control of enclosed ecosystems. To do this, a simulation-based controller was developed and implemented in a simulated greenhouse system. The role of the controller was to determine the setpoint path that would minimize the energy requirements for heating. It did this once a day, for the ensuing 24 hours, by simulating the greenhouse behavior in response to anticipated meteorological conditions. For each day a number of simulations were run for various setpoint scenarios and the most favorable scenario chosen. The greenhouse model that was used for these simulations was a neural network. The strategy used by the controller was based on the supposition that crops have a temperature integration capacity. Overall, the simulation-based controller allowed the greenhouse system to adapt itself to the anticipated disturbances and to behave more optimally than with a reference controller. It maintained high night temperatures when outside temperature and solar radiation intensity were anticipated to be low during the next day and vice-versa. This control approach reduced the average heating load by more than 7%. The results illustrate the adaptive capacity ofa simulation-based controller and provides a basis for the use of this approach in various other types of enclosed ecosystems. Keywords: Greenhouse control, simulation, artificial neural network. Afin d'explorer l'utilisation du controle par simulation pour des ecosystemes clos, un systeme de controle base sur la simulation (SCBS) a ete developpe et implante dans une serre simulee. Le role du SCBS etait de determiner une trajectoire de points de consigne pour Ie chauffage qui minimiseraient les besoins en chauffage. Cette trajectoire etait determinee une fois par jour, pour les 24 heures suivantes, en simulant avec un reseau neuronal Ie comportement de la serre en reponse aux conditions meteorologiques anticipees et a diverses trajectoires possibles. La strategie du SCBS etait basee sur la supposition que les plantes ont une capacite "d'integration de la temperature". Le SCBS a permis a la serre de s'adapter aux perturbations anticipees et de se comporter de fa~on plus optimale qu'avec un systeme de controle de reference. De hautes temperatures nocturnes ont ete maintenues lorsqu'il etait prevu que la temperature exterieure et les radiations solaires seraient basses durant Ie jour suivant; des temperatures basses etaient maintenues dans Ie cas contraire. L'approche utilisee par Ie SCBS a permis de reduire de plus de 7%, en moyenne, la charge de chauffage. Les resultats ont mis en relief la capacite d'adaptation d'un SCBS, et ont permis de jeter les bases d'un cadre methodologique pour la conception de tels systemes de controle, et pour leur implantation dans des ecosystemes clos.