|M.Sc Student||Waleed Agmail|
|Subject||MIMO H-infinity Control of Temperature and Humiditiy inside|
Greenhouses Cooled by Fogging Systems
|Department||Department of Civil and Environmental Engineering||Supervisors||Professor Linker Raphael|
|Dr. Avraham Arbel|
|Full Thesis text|
Despite the advantages of greenhouse fogging cooling systems over the more traditional cooling methods, such as sprinkling and pad-and-fan, this relatively new technique has never gained wide application in practice. The major reason for this is the lack of advanced algorithm for controlling the fogging and ventilation systems simultaneously.
Although some works were done in regard to fogging cooling systems, these were generally based on static modeling with simple on/off decision strategies. Moreover, most of these works considered temperature and humidity as two separate factors which do not necessarily have influence on each other. In this respect, these works have not focused on expressing the system as MIMO (Multiple-Input-Multiple-Output), which is more accurate and constitutes the key philosophy of the present research.
This work presents a MIMO approach for both modeling the non-linear physical system and the control problem. The modeling procedure relies on two differential equations for heat and mass transfer in the system which after linearization yields a 2x2 MIMO model which represents the physical behavior of the greenhouse. The control problem is based on defining two control inputs as the ventilation flow rate and the water flow rate which is regulated by water pressure in the fog lines, together with the outside enthalpy, humidity ratio and solar radiation as the disturbance inputs. This leads to a MIMO nominal transfer matrix which is initially shaped using two weighting matrices in order to meet the performance requirements, and then robustly stabilized in a one degree of freedom closed loop configuration using the LCF (Left Coprime Factorization) uncertainty formulation under the framework of the H∞ loop shaping optimization technique. This yields a robust 2x2 MIMO sub-optimal discrete controller with sampling time of 60 s which, based on simulations performed with the non-linear model, succeeds in maintaining the dry-bulb temperature and relative humidity inside the greenhouse very close to 28oC and 75% respectively, despite the presence of large parametric uncertainties and disturbances. This controller was then implemented in a small experimental greenhouse equipped with variable-speed extracting fans and two fogging lines in which the water pressure could be adjusted as desired. After a slight change in the controller gain in the second channel the controller was able to maintain the dry-bulb temperature and relative humidity within ±2oC and ±10% of their respective set points.