restriction of this technique is that the number of parameters to be estimated has to
be not to large compared with available measurement data.
All the aforementioned models deal with the nonlinearities of the greenhouse
system by using unconventional techniques to adjust the parameters of the model.
However, many researches choose linear model techniques, attempting to formulate an easier model, easily programmable in current greenhouse climate
controllers. This choice deals with many problems, taking into consideration
several restrictions have prove to work properly (Singh et al. 2006).
Many parametric models have been proposed, they usually are polynomial
equations which involve the main variables of phenomena (temperature, humidity,
among others) and many coefficients which are obtained by different identification
methods. The well-known off-line least squares method was used by Bennis et al.
(2008), input–output signals sampled every minute during a day were used to build
the model. The model also involves perturbation such as external humidity and
temperature, radiation, and wind speed. Previous works opt for on-line identification, arguing that changing the operation conditions in the greenhouse requires
re-tuning of the controller parameters in order to achieve optimum performance.
However, fewer variables are taking into account in this methodology because the
computational power required increases according with the number of variables to
estimate (Arvanitis et al. 2000).
Despite the favorable results obtained by the aforementioned models, many
authors argue that a more detailed knowledge about phenomena is required in
order to increase the performance of controller and thus to meet objectives of
modern agriculture (Katsoulas et al. 2007). Some works deal the problem of model
plant behavior and it influence on the greenhouse microclimate. These behaviors
could be related with plant growth, transpiration, or photosynthesis (Boulard and
Wang 2000).
Other works focus in order to understand de ventilation phenomena inside the
greenhouse by models with consider the heterogeneity of the system and how the
different segments of the atmosphere interact with plan canopy or external environment (Abdel-Ghany and Al-Helal 2011; Dayan et al. 2004). Computational
fluid dynamics (CFD) also serves to study ventilation, its effects on inside temperature and humidity, and how it is affected by external wind direction (Bournet
and Boulard 2010; Kittas and Bartzanas 2007). Experimental work focused in
understanding how the ventilation rate modifies the physiological state of the plan
was conducted. Finally based on these studies, different methodologies were
purposed to design the greenhouse geometry, the correct orientation, and even the
required equipment in order to achieve a desirable yield and quality (Nebbali et al.
2012; Vanthoor et al. 2011). Also, software which use all the aforementioned
models were developed in order to provide a tool that could predict the greenhouse
behavior under certain weather characteristics previously its installation, equipment available, and greenhouse materials could also be included in this web-base
application (Fitz-Rodríguez et al. 2010).
13 Instrumentation and Control to Improve the Crop Yield
391
be not to large compared with available measurement data.
All the aforementioned models deal with the nonlinearities of the greenhouse
system by using unconventional techniques to adjust the parameters of the model.
However, many researches choose linear model techniques, attempting to formulate an easier model, easily programmable in current greenhouse climate
controllers. This choice deals with many problems, taking into consideration
several restrictions have prove to work properly (Singh et al. 2006).
Many parametric models have been proposed, they usually are polynomial
equations which involve the main variables of phenomena (temperature, humidity,
among others) and many coefficients which are obtained by different identification
methods. The well-known off-line least squares method was used by Bennis et al.
(2008), input–output signals sampled every minute during a day were used to build
the model. The model also involves perturbation such as external humidity and
temperature, radiation, and wind speed. Previous works opt for on-line identification, arguing that changing the operation conditions in the greenhouse requires
re-tuning of the controller parameters in order to achieve optimum performance.
However, fewer variables are taking into account in this methodology because the
computational power required increases according with the number of variables to
estimate (Arvanitis et al. 2000).
Despite the favorable results obtained by the aforementioned models, many
authors argue that a more detailed knowledge about phenomena is required in
order to increase the performance of controller and thus to meet objectives of
modern agriculture (Katsoulas et al. 2007). Some works deal the problem of model
plant behavior and it influence on the greenhouse microclimate. These behaviors
could be related with plant growth, transpiration, or photosynthesis (Boulard and
Wang 2000).
Other works focus in order to understand de ventilation phenomena inside the
greenhouse by models with consider the heterogeneity of the system and how the
different segments of the atmosphere interact with plan canopy or external environment (Abdel-Ghany and Al-Helal 2011; Dayan et al. 2004). Computational
fluid dynamics (CFD) also serves to study ventilation, its effects on inside temperature and humidity, and how it is affected by external wind direction (Bournet
and Boulard 2010; Kittas and Bartzanas 2007). Experimental work focused in
understanding how the ventilation rate modifies the physiological state of the plan
was conducted. Finally based on these studies, different methodologies were
purposed to design the greenhouse geometry, the correct orientation, and even the
required equipment in order to achieve a desirable yield and quality (Nebbali et al.
2012; Vanthoor et al. 2011). Also, software which use all the aforementioned
models were developed in order to provide a tool that could predict the greenhouse
behavior under certain weather characteristics previously its installation, equipment available, and greenhouse materials could also be included in this web-base
application (Fitz-Rodríguez et al. 2010).
13 Instrumentation and Control to Improve the Crop Yield
391