We have presented the application of an automated
greenhouse system using NI Embedded Systems
Technology. The system is instrumented with sensors and
actuators and controlled by an embedded cascaded PID
algorithm in order to provide accurate control of the air
temperature within the greenhouse. A simulation model
of a subsystem has also been developed and shows good
agreement to the experimental results. Further modeling
work is required to describe accurately the whole
system’s dynamics and thus design more advanced
control algorithms to ensure minimum energy
consumption and accurate control of the environmental
conditions.
9 trang |
Chia sẻ: huongthu9 | Lượt xem: 502 | Lượt tải: 0
Bạn đang xem nội dung tài liệu Environmental Control of a Greenhouse System Using NI Embedded Systems Technology, để tải tài liệu về máy bạn click vào nút DOWNLOAD ở trên
Environmental Control of a Greenhouse System
Using NI Embedded Systems Technology
Christakis Papageorgiou, Ahmed Sadriwala, Mohammed Almoalem, and Conor Sheedy
School of Engineering, Bahrain Polytechnic, Kingdom of Bahrain
Email: c.papageorgiou@polytechnic.bh
Andre Hajjar
National Instruments (NI)
Email: Andre.Hajjar@ni.com
Abstract—This paper presents the application of an
automated environmental control system for a prototype
greenhouse system using commercial embedded systems
technology. The prototype greenhouse system was developed
and instrumented with appropriate sensors to measure
various environmental variables like the temperature, the
light intensity, the soil moisture, the air humidity and CO2
concentration. These measurements are provided to the
control algorithm which is implemented on a commercial
embedded system and manipulates various actuators like, a
heating and cooling actuator, fans, lights, irrigation system,
and louvers in order to achieve the desired set-points, as
specified by the user through a Human-Machine Interface
implemented in LabView software. Certain aspects of the
greenhouse dynamics have been modeled in
Matlab/Simulink using nonlinear differential equations and
the simulation model has been validated against
experimental data, showing good agreement between the
simulation and the experimental data. The purpose of this
work is to enhance research related to the accurate
environmental control of greenhouse systems in order to
minimize energy and water consumption and to develop a
robust educational platform for teaching control system
design, analysis, instrumentation and embedded systems
development at the Engineering School of Bahrain
Polytechnic.
Index Terms—environmental control, automated
greenhouse system, embedded system, temperature
regulation
I. INTRODUCTION
A. Environmental Control of Greenhouse Systems
The environmental control of a greenhouse implies the
regulation of day and night air temperatures, the relative
humidity and the carbon dioxide levels to ensure optimal
plant growth. Heat, water vapor and carbon dioxide are
transferred in and out of the greenhouse space in order to
maintain the required set-points of temperature, relative
humidity and carbon dioxide concentration. Heat is
transferred by conduction, convection and radiation and
various mass transfer processes are also occurring
utilizing fans and louvers resulting in complex flows that
Manuscript received July 22, 2015; revised November 2, 2015.
involve both heat and mass transfer. A good introductory
reference detailing the basic functions of a greenhouse
system is found in [1]. A more detailed account of the
functionality of an automated greenhouse system is
presented in [2]. The authors discuss the various
functionalities of a greenhouse such as light intensity
control, heating control, cooling control, air circulation
control and humidity control. Various active and passive
actuation devices are presented in order to achieve the
desired regulation effect. A brief discussion on closed-
loop (feedback) control is given, detailing some
requirements on sensing environmental variables and on
control algorithm implementation. Interestingly, the
authors present their analysis based on the application of
a greenhouse automated system for the extreme Alaskan
weather. They quote: ``By optimizing light, temperature
and humidity, in conjunction with the proper fertilization,
watering and selection of adapted varieties, an endless
array of growing opportunities await the Alaska
greenhouse gardener and commercial producer’’. Moving
towards the more moderate Mediterranean climate, the
authors in [3] present an analysis of the most important
functionalities of an automated greenhouse system,
namely the temperature and relative humidity control.
The authors provide indicative set-point values for these
variables that favor plant growth and based on the
climograph information for a given location, they
determine the levels of cooling or heating required to
maintain these set-point values. The climograph contains
information regarding the mean solar radiation and mean
air temperature for a given location all around the year. It
constitutes the starting point in identifying the actuation
requirements for an automated greenhouse system.
Special attention is given to energy efficiency and
sustainability. The authors focus on providing favorable
conditions for plant growth during the hot periods of the
Mediterranean climate while using energy efficient
processes like ventilation, shading, evaporative cooling
and effective insulation. Similar design requirements are
addressed in [4], where the authors present the
application of an automated greenhouse system for the
production of tomatoes.
331©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
doi: 10.18178/joace.4.5.331-339
B. Modelling of Greenhouse Systems
The automation of a greenhouse implies the
implementation of a closed-loop (feedback) control
algorithm, from the simplest on/off strategies employing
thermostats to the most advanced PID and gain-scheduled
controllers requiring accurate sensor measurements and
algorithm development on embedded systems. The
advantage of the implementation of a more advanced
control algorithm is the ability to account for optimal
design and operation requirements, like minimum energy
control and accurate regulation in the presence of external
weather disturbances and uncertainty in modeling
accurately the biological processes characterizing the
growth of plants. The design of an advanced control
algorithm requires a good knowledge of the open-loop
system dynamics, in our case, the greenhouse system.
The process of acquiring knowledge and representing the
open-loop dynamics of a system is called physical and
mathematical modeling. The objective is to obtain a set of
mathematical equations relating the inputs to the outputs
of the greenhouse system. The outputs are chosen to be
the environmental variables that we wish to regulate, like
the temperature of the air, the relative humidity, the soil
moisture, the light intensity and any other environmental
state that we can measure with an appropriate sensing
device. The inputs of the system can be the power of the
heating or cooling actuators, the power of the fans, the
opening of the ventilation ducts, the coverage of the
shading system, etc. These mathematical equations that
describe the system dynamics are usually very
complicated and nonlinear for a given physical system
like the greenhouse system. For this reason, it is more
convenient to represent these equations in the format of a
simulation model in appropriate simulation software. This
implies that the nonlinear, continuous time equations are
discretized and solved numerically using a computer. The
available computing power these days provides for very
accurate solutions for the majority of physical systems,
and most definitely for greenhouse systems, whose
dynamics evolve over relatively slow times due to their
large thermal inertia. The formulation of the
mathematical equations themselves is quite challenging
due to the distributed nature of the thermal system of a
greenhouse. This means that environmental variables like
temperature and relative humidity vary not only
temporally (over time) but also spatially (over distance).
On top of that, the governing physical processes are quire
complex, involving highly nonlinear processes like heat
transfer through convection and radiation, forced fan
flows, ventilation flows, transpiration of plants,
photosynthesis, condensation, chemical reactions and
external disturbances due to external weather changes.
The task of modeling the greenhouse dynamics remains
challenging depending on the level of model complexity
required.
Whether the distributed system modeling approach or
the lumped capacitance modeling approach is used, the
underlying physical laws of conservation of energy and
conservation of mass are the starting point of the
modeling process. The authors in [5] have used the
lumped capacitance modeling approach to derive
differential equations in state-space form by applying the
principle of conservation of energy governing the heat
transfer processes occurring in a typical greenhouse
system. The authors in [6] have used the thermal network
approach to model the dynamics of a
solarium/greenhouse system. The thermal network
approach has strong links to the electrical network
approach, where various nodes are chosen characterized
by their respective temperatures. Subsequently, the law of
conservation of energy is applied at each node to
determine the balance between the internal energy at the
node and the energies transferred in and out of the node.
Similar to the electrical network approach, the stored
internal energy of the node is influenced by the thermal
capacitance of the node and the amount of energy
transferred in and out of the node is influenced by the
thermal conductance of the paths between adjacent nodes.
The authors in [7] have utilized a dedicated software
modeling package (TRNSYS) to derive a thermal model
for an experimental greenhouse in Nepal. A very
informative account of the modeling process adopted for
the characterization of greenhouse dynamics can be found
in [8]. The authors present a detailed analysis of the
various physical processes affecting the greenhouse
dynamics, like the external solar radiation, the heat
transfer processes due to convection and conduction, the
air flows due to leakages and natural ventilation, the
water vapor fluxes and the canopy transpiration. The
modeling process follows the lumped capacitance
modeling approach where the greenhouse system is
separated into a finite number of subsystems each
characterized by its temperature. One of the key
assumptions is that all subsystems are considered
homogeneous, which implies that they have a uniform
temperature and an average thermal capacity can be used
for each subsystem.
C. Project Objectives
The objectives of the greenhouse project are:
To develop a prototype greenhouse system that is
equipped with appropriate instrumentation in
terms of sensors, actuators and control
hardware/software,
To design control algorithms to regulate the
environmental variables within the greenhouse
system,
To implement the control algorithms on a robust
embedded platform and to construct a Human-
Machine Interface between the operator and the
greenhouse system,
To apply physical and mathematical modeling
techniques for the construction of a high-fidelity
simulation model for the greenhouse system,
To investigate the design of advanced control
algorithms for minimum energy consumption,
To investigate the design of more efficient passive
and active actuators for cooling and heating,
To utilize the greenhouse system as an educational
platform for the design and implementation of
control algorithms.
332©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
Objective 1 has been achieved with the construction of
the prototype greenhouse system, fully instrumented with
environmental sensors, actuators and control hardware.
Objective 2 has been partially achieved with the design of
a control system for the regulation of the greenhouse air
temperature. Further work is required to design control
systems for the regulation of other environmental
variables like the relative humidity, the light intensity, the
soil moisture and the carbon dioxide concentration.
Objective 3 has been fully achieved with the design and
implementation of the control algorithms on the robust
embedded platform NI CompactRIO
( and the design of a
Human-Machine Interface using the LabView software
( Objective 4 has been
partially achieved with the partial modeling and
validation of the heating actuator of the greenhouse
system. A simulation model of the heating actuator
subsystem has been constructed using the
Simulink/Matlab software. Further work is required to
model and validate the complete greenhouse system
dynamics. Objectives 5, 6 and 7 have not been addressed
at this stage and will be considered in future research and
development work.
D. Layout of the Paper
The remainder of the paper will consist of two main
sections, the Prototype Greenhouse System section and
the Experimental Results section.
The first section will address the hardware and
software development for the system in more detail. A
subsection will be devoted to the description of the
functionality of the system, a subsection will be devoted
to the presentation of the hardware and software
components and a third subsection will focus on the
presentation of the modeling work related to the heating
actuator subsystem.
The second section will focus on the presentation of
the experimental results from the testing of the heating
actuator subsystem in order to validate the simulation
model. Also, a subsection is devoted to the presentation
and analysis of the closed-loop experimental results from
the operation of the greenhouse system.
Finally, the conclusions will be presented along with
directions for future research and development work.
II. THE PROTOTYPE GREENHOUSE SYSTEM
A. System Functionality
Figure 1. The prototype greenhouse system
In this section, we present the hardware functionality
of the prototype greenhouse system, which is described in
more detail in [9]. The prototype greenhouse system is
shown in Fig. 1. From the far left we can identify the
embedded system NI CompactRIO, the water tank with
the yellow lid, the heating actuator, the greenhouse and
on the far right the cooling actuator and the HMI (host
computer). The following control systems are
implemented.
1) Greenhouse temperature regulation
The greenhouse air temperature is regulated by a
cascaded PID controller architecture. There exist two
inner PID controllers manipulating the PWM signals to
the heating elements and the cooling Peltiers respectively
in order to maintain the desired air temperatures within
the heating and cooling actuators.
There also exist two outer PID controllers that
manipulate the PWM signals of the fans for the heating
and cooling actuators respectively. The outer PID
controllers react to the error signal between the desired
and actual greenhouse temperature in order to induce the
flow of the right proportion of hot and cold air into the
greenhouse.
2) Regulation of light intensity
There exists capacity for a control system to regulate
the light intensity in the greenhouse. The control system
receives an input from the light sensor and manipulates
the actuator (Halogen Lights) in order to achieve the
desired set-point of light intensity. This system can also
operate on Manual Mode, with the user varying directly
the intensity of the halogen lights.
3) Regulation of soil moisture
There exists capacity for a control system to regulate
soil moisture content. The measurement is provided by an
electrical conductivity sensor that relates the conductivity
of the soil to the soil moisture. On Automatic Mode the
controller controls the Actuator (Water Pump) but on
Manual Mode the user will be able control the actuator
directly and supply the desired amount of water to the
plants.
4) Monitoring of other environmental conditions
Besides the regulation of the greenhouse air
temperature, light intensity and soil moisture, the system
monitors and displays the following environmental
conditions:
Relative Humidity inside the Greenhouse.
Relative Humidity of the Environment.
CO2 concentration inside the Greenhouse.
Each environmental condition is displayed in the form
of a real-time graph using the LabView front panel
application.
5) Automated doors
The greenhouse has an automated door which opens
when a person comes close to the proximity sensor and
closes when the person has completely entered the
greenhouse. The reason for an automated door is to
ensure the door remains closed so that heat losses are
minimized and the temperature control system has
smaller external disturbances to cope with.
333©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
6) Human-Machine Interface (HMI)
There is a LabView HMI which provides both data-
entering and data-receiving functionality and allows the
user to communicate efficiently with the system by
receiving real-time information regarding the status of the
system and also to enter the set-points for the various
control systems. The HMI allows both automatic control
operation and manual control operation.
7) Safety features of the system
The HMI interface is secure and restricts unauthorized
personnel to change the settings and parameters of the
system, since at every step of the program the user is
asked to enter a password.
8) Other features of the system that add flexibility
The user is able to control the angle of the vents
(louvers)and the power of the ventilation fan under
manual mode.
There is a data logging function with which the
user can save real-time data.
B. Hardware and Software Development
The following NI hardware and software products have
been utilized for the implementation of the prototype
greenhouse system.
1) Software
a) LabView FPGA Module
b) LabView Real-Time Module
c) NI-RIO Driver
d) LabView Control Design Toolkit
e) NI LabView PID and Fuzzy Logic Toolkit
f) LabView MathScript RT Module
g) LabView Control Design and Simulation Module
h) NI MAX (National instrument measurement and
Automation Explorer)
2) Hardware
a) NI CompactRIO
b) Power supply NI PS-15
Below we present the sensor and actuator modules:
NI 9203: This is a module that receives a measured
current signal in the range -20 to 20 mA. This module is
used to receive the measurement of the temperature
sensor and the relative humidity sensors. This module
provides a 16 bit resolution.
NI 9215: This is a module that receives a measured
voltage signal in the range -10 to 10 V. This module is
used to receive the measurements of the light sensor, the
soil moisture sensor and the capacitive humidity sensor.
This module provides an adequate resolution and a high
sampling rate which makes it suitable for the light
intensity control system.
NI 9401: This is a digital input/output module that
receives a measured digital voltage signal of 5 V. This
module is used to receive the measurements from all the
digital sensors which are contact switches, proximity
sensors and limit switches. The importance of this
module is its bi-directionality which allows it to be used
both as an input and output digital signal module. It is
used for producing the PWM signal for all the actuators,
like the PTC (positive temperature coefficient) heaters,
the cooling Peltiers, the fans and the pumps.
NI 9219: This module receives signals from the
thermocouples which are placed in various parts of the
system. The thermocouples are wired directly to it and it
reads the voltages at a resolution of 24 bits. This module
is a universal Analog input which can read various
signals like voltage, current and temperature.
3) Embedded system architecture (Fig. 2)
a) CompactRIO
The CompactRIO is an embedded platform that allows
the real-time implementation of algorithms. Due to the
FPGA architecture it allows parallel processing and data
acquisition. Its modular architecture allows the
simultaneous use of multiple input and output modules.
The modules selected for our greenhouse system are
compatible to a large variety of industrial sensors which
makes it straightforward for data acquisition and analysis.
When using the CompactRIO, it is straightforward to
change modules efficiently and due to its reconfigurable
FPGA and Real-Time Processor, it allows for rapid
prototyping of new engineering systems. The FPGA
architecture also allows for parallel processing and the
real-time Processor can work in stand-alone mode
without being connected to the host program.
b) Host computer with LabView
In this project a laptop with a LabView installation is
used as the Host. The Host provides the environment for
a LabView application to receive and send data to the
CompactRIO.
c) LabView application
In this project, 3 main VIs had to be created. The first
one is the Host.vi which is a LabView Program that
allows the user to interact with the greenhouse system
(Human Machine Interface). The second one is the
Target.vi which runs on the Real-Time Processor. This
application provides communication between the Host
and the FPGA. The third one is the FPGA.vi which runs
on the FPGA and it provides data acquisition and
interfacing with the sensors and actuators. All VIs
communicate with each other to achieve accurate control
of the greenhouse system. The benefit of using LabView
is its graphical programming capability and the
transparency it provides to the software engineer. Many
built-in functions of LabView were used like data logging,
simulation kits and control kits in order to design the
application for the greenhouse system. Since the project
is focused on automation and control system design, the
control system design toolkit was used extensively.
334©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
Controller NI cRIO-9025
Chassis cRIO-9118
Modules 5 x NI 9401
Module 1 x NI 9215
Module 1 x NI 9203
Module 2 x NI 9219
Figure 2. Architecture of the embedded system for the greenhouse control
C. Mathematical Modelling of the Heating Actuator
This section presents the mathematical modeling of the
heating actuator subsystem that is used to provide heat to
the greenhouse. The method that was adopted will be
briefly explained, while the key equations will be
discussed. The operation of the heating actuator
subsystem relies on the operation of the PTC (Positive
Temperature Coefficient) heating elements that convert
the electrical power into heat. This heat is stored in a heat
sink and transferred by forced and natural convection to
the air around and above the heat sink. The operation of
the two fans transfers the hot air into the greenhouse in
order to increase the greenhouse air temperature.
1) Lumped capacitance modelling methodology
Fig. 3 shows a schematic diagram of the heating
actuator subsystem. The two inputs of the system are the
duty cycle of the heating elements (T% Heat) PWM signal
and the duty cycle of the fans (T% Fans) PWM signal. The
outputs are the temperature of the sink (Tsink) and the
temperature of air (Tair) within the heating actuator. The
temperature of the input air (Tin) and the environment
temperature (Tenv) are external inputs to the system. All
temperatures are in degrees Celsius.
We use the lumped capacitance modeling methodology,
described by the following equation, in order to model
the heating actuator subsystem:
( )i n o u t
d T
P P C
d t
where i nP
is the power input in Watts, o u tP is the
power output in Watts, C is the thermal capacitance in
J/K, T is the temperature in K and t is the time in
seconds.
Figure 3. Schematic of the heating actuator subsystem
This equation is basically a statement of the
conservation of energy, stating that the rate of increase of
the internal energy of the system is equal to the total
power supplied to the system minus the total power lost
by the system.
The above equation is used twice, once for the heat
sink and once for the air surrounding the heat sink. We
assume that the temperature of the PTC heating elements
is uniform and equal to the heat sink temperature and also
that the air temperature is uniform throughout the
enclosure of the heating actuator. Two heat transfer
processes are considered to transfer heat from the heat
sink to the air, namely those of convection and radiation.
The air will gain heat from the heat sink due to
convection and radiation, and will lose heat due to the
flow of air out of the heating actuator subsystem, and also
due to conduction losses through the walls of the actuator.
Thus, we have the following two equations:
e l e c t r i c a l c o n v e c t i o n r a d i a t i o n
s i n k
s i n k
P P P
d T
C
d t
335©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
c o n v e c t i o n r a d i a t i o n c o n d u c t i o n
a i r
a i r f l o w a i r
P P P
d T
P C
d t
The various power components are calculated as
follows.
e l e c t r i c a lP is the total electrical power input
supplied by the PTC heating elements. It is determined by
multiplying the squared peak voltage, by the duty cycle
and dividing by the resistance of the heating elements.
2
0 %
( )
e l e c t r i c a l
s i n k
V
P T
R T
The temperature of the heating elements affects their
resistance. Being PTC implies that the resistance
increases as the temperature of the heating element
increases. The relationship between temperature and
resistance was found experimentally and it is shown in
Fig. 4.
c o n v e c t i o nP is assumed to be due to natural
convection when the fans are turned off, and due to
forced convection when the fans are turned on. The
equations for calculating the heat transfer due to both
forced and natural convection were taken from [10], and
are of the form:
s i n
A ? )c o n v e c t i a io n k rP h T T
Figure 4.
Nonlinear relationship between the temperature of the PTC
heating element and its resistance
The convection coefficient h and the convection area
A change accordingly to the forced or natural
convection situations.
It is important to note here that the
convection coefficient h is a highly nonlinear function
of the flow characteristics and the geometry
characteristics of the heat sink fins, [11].
r a d i a t i o nP is obtained by the analytical equation
below:
4 4ε A σ ( )r a d i a t i o n s i n k a i rP T T
where is the emissivity of the body, A
is the surface
area that radiates heat, and
is the Stefan-Boltzmann
Constant (equal to
8
2 4
W
5 . 6 7 0 3 1 0
m K
) , [12] .
c o n d u c t i o nP
is obtained from the following equation:
( )
c o n d u c t i o n
a i r e n vP k A
d
T T
where k is the thermal conductivity coefficient of the
acrylic material that is insulating the heating actuator
space, A is the surface area of that material and d is the
thickness of that material, [12].
a i r f l o wP
is the rate at which heat is transferred out of
the actuator space due to the fan-induced air flow from
the inlet to the outlet of the heating actuator subsystem. It
is represented by the following equation:
? )
a i r
a i r i no u t a i r
a i r f l o wP pu c T T
where a i r is the density of air, a i rpc is the specific heat
capacity of air, and o u tu
is the volumetric flow rate
of air.
The model accounts for the variation of the properties of
air as a function of the air temperature using appropriate
look-up tables.
The volumetric flow rate is also a
nonlinear function of the duty cycle of the PWM signal
that drives the fans. Fig.
5
presents this nonlinear
relationship that was obtained experimentally. It was
noted that due to static friction effects, the fans only start
rotating after the duty cycle
of the PWM signal exceeds
20%. The sink thermal capacitance,
s i n kC ,
is obtained by
multiplying the mass of the heat sink with
the specific
heat capacity of the sink’s material.
An adjustment of
s i n kC will be required during the validation phase since
the heat sink is designed to act as a heat conductor and
not as a heat storage device.
Figure 5. The variation of the volumetric flow rate of the fans as a
function of the duty cycle of the PWM input signal
Therefore, the actual heat sink thermal capacitance
maybe less than the theoretical value calculated above.
a i rC is obtained by multiplying the specific heat
capacity of air by its mass. The mass of air is found by
multiplying its density by the approximate volume of the
air in the heating actuator subsystem. Once all the
336©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
equations were formulated, the simulation model was
constructed in Simulink
( and all
the equations were implemented. The simulation model
will be validated against experimental data below.
III. EXPERIMENTAL RESULTS
A. Validation of the Simulation Model
The simulation model of the heating actuator was
validated against experimental data obtained from
numerous experiments under different testing conditions.
Initial results were very promising but further modeling
work is required in order to enhance the fidelity of the
simulation model. The work related to the modeling of
the heating actuator subsystem is presented in [13].
The experimental testing for the heating actuator
focused mainly on the variation of the PWM signal to the
heating elements and the measurement of the resulting
temperatures of the heat sink and the surrounding air as a
function of time. The fans were left switched off. This
methodology aimed to test independently and
sequentially the effects of the two inputs and provide
appropriate experimental data in order to make rational
adjustments to the simulation model. The objective in
making adjustments to the simulation model is to
minimize the error in the temperature responses between
the simulated and the experimental data.
An example of the validation exercise is shown in Fig.
6. The top subplot represents the PWM signal for the fans,
which is zero therefore the fans are left switched–off. The
second subplot represents the PWM signal for the heating
elements and this is chosen as a square pulse signal of
varied amplitude in order to investigate both the heating
and natural cooling response of the system. The
experiment lasts for 30 minutes and during this time the
temperatures of the heat sink and the surrounding air are
measured and recorded. These are presented in the 3
rd
and
4
th
subplots respectively. From an initial look the results
look satisfactory and the same trends are observed both in
the simulated and the experimental data. The differences
in the simulated and experimental temperatures are of the
order of 10% for the heat sink temperature and of the
order of 12% for the air temperature.
From a control designer’s point of view, the agreement
between the model and the real system shown in Fig. 6 is
more than satisfactory for the design of a good feedback
control system. Steady-state errors can be accommodated
using integral action in the controller.
Further experimental testing is required with the
inclusion of the fans input. This will be conducted at a
future stage and future modifications in the model are
likely to occur in order to capture the real dynamics
occurring with the variation of both the inputs of the
system.
B. Closed-loop Experimental Testing
In this section we will present the results from the
closed-loop experimental testing of the greenhouse air
temperature control system. The air temperature within
the greenhouse is considered as the highest priority
environmental condition that needs to be regulated
accurately. The control system block diagram is shown in
Fig. 7. The functionality of the closed-loop system is that
of cascaded PID control systems. The objective is to
regulate the greenhouse temperature by blowing the
correct proportion of hot and cool air in the greenhouse.
The hot and cool air is generated independently in the
respective heating and cooling actuators. Therefore, the
inner PID controllers, one for the heating actuator and
one for the cooling actuator are responsible for
maintaining the desired air temperature within the
actuator subsystems. These inner PID controllers
manipulate directly the PWM signals to the PTC heating
elements and the cooling Peltiers respectively. It is
evident from the block diagram that the reference
temperature for the heating actuator air is 50 degrees
Celsius and the reference temperature for the cooling
actuator air is 18 degrees Celsius.
Figure 6. Comparison of simulated and experimental data from the
testing of the heating actuator subsystem
Figure 7.
Block diagram of the closed-loop system for the regulation
of the greenhouse temperature
The outer loop PID controllers manipulate the fans
within the heating and cooling actuators respectively
337©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
based on the error signal between the reference
greenhouse air temperature and the measured greenhouse
temperature. In this way the correct proportion of hot and
cold air is sent into the greenhouse in order to maintain
the air temperature at its desired set-point.
Figure 8. Regulation of the greenhouse temperature
An example of the operation of the closed-loop system
is shown in Fig. 8. The desired set-point for the
greenhouse temperature is started at 25 degrees, then put
at 28 degrees, then put back at 25 degrees, then put at 22
degrees and finally brought back at 25 degrees. We
assume that the range of temperatures between 22-28
degrees Celsius is a favorable temperature range for the
growth of most plants. The 1
st
subplot in Fig. 8 shows the
good tracking of the set-point. The 2
nd
subplot presents
the PWM signals for the heating elements and the cooling
Peltiers required to maintain the heating actuator and
cooling actuator temperatures at 50 degrees and 18
degrees Celsius respectively. The temperatures of the
heating and cooling actuators are presented in the 3
rd
subplot, where it is clearly shown that the regulation is
very accurate, even in the presence of disturbances like
the starting of the fans in the two actuators. The 4
th
subplot presents two external disturbance signals, the
vents and the fan. These disturbance signals are used
manually by the operator in order to demonstrate the
good disturbance rejection of the control system. It is
evident that the fan disturbance is more severe, but even
in that case the control system manages to reject the
disturbance and regulate accurately the greenhouse
temperature by manipulating accordingly the fans in the
heating and cooling actuators. The 5
th
subplot presents the
PWM signals for the fans in the heating and cooling
actuators. These are the control signals that are
manipulated by the outer-loop PID controllers and are
responsible for maintaining the desired greenhouse
temperature. We can observe, for example, that around
500 seconds the set-point is put at 28 degrees, i.e. 3
degrees higher than the previous value. The control
system reacts to this demand by increasing the PWM
signal to the fan of the heating actuator in order to send
more hot air in the greenhouse and thus increase its
temperature. Similarly, around 1100 seconds, we demand
a decrease in the greenhouse temperature by 3 degrees,
from 25 to 22. The control system reacts to this demand
by increasing the PWM signal to the fan of the cooling
actuator in order to send more cold air into the
greenhouse and thus reduce its temperature.
The parameters of the PID controllers, both inner and
outer loop were designed after successive iterations of
testing and experimentation, due to the lack of an
accurate simulation model for the whole greenhouse
system. In the future, we will develop further the
simulation model for the greenhouse system in order to
facilitate the application of more systematic control
design techniques and thus reduce substantially the
control development time.
IV. CONCLUSIONS
We have presented the application of an automated
greenhouse system using NI Embedded Systems
Technology. The system is instrumented with sensors and
actuators and controlled by an embedded cascaded PID
algorithm in order to provide accurate control of the air
temperature within the greenhouse. A simulation model
of a subsystem has also been developed and shows good
agreement to the experimental results. Further modeling
work is required to describe accurately the whole
system’s dynamics and thus design more advanced
control algorithms to ensure minimum energy
consumption and accurate control of the environmental
conditions.
ACKNOWLEDGMENT
The authors wish to thank Mr. David Krause,
Programme Manager of the Electronics Programme at
Bahrain Polytechnic, for his initial inspiration in
constructing, instrumenting and controlling a greenhouse
model prototype as part of a student final-year project. Le
REFERENCES
[1] W. J. Roberts, “Environmental control of greenhouses,” Technical
Report, CCEA, Centre for Controlled Environment Agriculture,
Cook College, Rutgers University, 2005.
[2] W. Vandre and L. Clayton, “Controlling the greenhouse
environment,” Technical Report, Cooperative Extension Service,
University of Alaska Fairbanks, 2003.
[3] C. Kittas, N. Katsoulas, and T. Bartzanas, “Greenhouse climate
control in mediterranean greenhouses,” Cuadernos de Estudios
Agroalimentarios, pp. 89-114, CEA03, July 2012.
[4] R. G. Snyder, “Environmental control of greenhouse tomatoes,”
Technical Report, Extension Service, Mississippi State University,
2003.
[5] N. Bibi-Triki, S. Bendimerat, A. Chermiti, T. Mahdjoub, B.
Draoui, and A. Abene, “Modelling, characterization and analysis
of the dynamic behavior of heat transfers through polyethelene
and glass walls of greenhouses,” Physics Procedia, vol. 21, pp.
67-74, 2011.
338©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
[6] D. Bastien and A. K. Athienitis, “A control algorithm for optimal
energy performance of a solarium/greenhouse with combined
interior and exterior motorized shading,” Physics Procedia, vol.
30, pp. 995-1005, 2012.
[7] S. Candy, G. Moore, and P. Freere, “Design and modelling of a
greenhouse for a remote region in Nepal,” Procedia Engineering,
vol. 49, pp. 152-160, 2012.
[8] M. Binotto, “Greenhouse climate model,” M.S. thesis, Reykjavík
University, 2012.
[9] A. Sadriwala, “Greenhouse automation systems,” Technical
Report, Bahrain Polytechnic Engineering School, 2014.
[10] J. G. Corominas, “Heat sink analytical modelling,” M.S. thesis,
Universitat Politècnica de Catalunya, École Supérieured’
Electricité, 2011.
[11] D. Roncati. (n.d.). Iterative calculation of the heat transfer
coefficient. [Online]. Available:
[12] R. Nave. (April 4, 2015). (n.d.). Heat Conduction. [Online].
Available:
astr.gsu.edu/hbase/thermo/heatcond.html
[13] M. Almoalem, “Greenhouse modelling project,” Technical Report,
Bahrain Polytechnic Engineering School, 2015.
Christakis Papageorgiou was born in
Limassol, Cyprus in 1975. He studied at the
University of Cambridge, UK, between 1995-
1999 for an MEng in the Electrical and
Information Sciences Tripos. He continued his
studies at the same University for a PhD in
Control Systems, which he completed in
February 2004.He worked in the Control Lab
as a Research Associate at the University of
Cambridge, until 2006, where he carried-out
research on a novel mechanical element, the inerter. Between January
2005 and January 2006, he was also a Research Associate in the
Department of Electrical and Computer Engineering at the University of
Cyprus, Nicosia, Cyprus, working on a project focused on the utilization
of the inerter for the design of optimal passive vehicle suspensions. He
has conducted research on a European-funded project on the clearance
of flight control laws using optimization for civil aviation aircraft while
being a researcher at Linkoping University in Sweden between 2007
and 2010.He is currently the Head of the Engineering School at Bahrain
Polytechnic, responsible for the delivery of the BEngTech degree
through the implementation of problem-based learning. His research
interests include control and modeling of environmental systems, robust
and optimal control with linear matrix inequalities, analysis and design
of control laws for aeroservoelastic systems, vibration suppression and
rejection of structural vibrations, robustness analysis and synthesis of
nonlinear flight control laws, experimental testing of passive
mechanical networks for vehicle suspensions, and the implementation
of Problem-Based Learning for the delivery of an Engineering
curriculum.
Ahmed Najmuddin Akber Ali Sadriwala
was born in India 1990, holding a Bahrain
Nationality. After graduating from the Indian
School in Bahrain, he joined Bahrain
Polytechnic in 2010 to start his BEngTech
degree in Electronic Engineering. During his
time in Bahrain Polytechnic, Ahmed has done
many projects like Portable Halogen Strobe
Light, RC Car, Elevator Programming and
Electronic Dice using Logics. His Graduation
Project was Greenhouse Automation using the MBED microcontroller.
The Project won 2nd price in the Garden Show Exhibition in Bahrain
(BIGS 2015). Later the Project was modified under the supervision of
Dr Christakis Papageorgiou to use industrial embedded systems from NI
to implement a Control System to regulate different environmental
variables of the Greenhouse. Regarding Industrial Experience, he has
worked as an intern in Steel United, Bahrain as Maintenance trainee and
in Bahrain Polytechnic as a Technician. He is currently working in
Pipeline Technologies and Services as a Supervisor of the Assembly
Department.
Mohammed Almoalem
was born in Manama,
Bahrain in 1994. After graduating from
Sheikh Abdulaziz Secondary School with a
grade of 95.8% in Mathematics and Physics,
he joined Bahrain Polytechnic in 2012 to start
his BEngTech in Mechanical Engineering.
Ever since, Mohammed has
completed several
projects as part of his studies such as
designing, analyzing and manufacturing a
vehicle suspension system and chassis.
Between February and June 2015, he
completed Modelling the Heating
Actuator of a Greenhouse prototype
system. His academic achievements were noticeable and
he received the
Dean’s award of the Highest Achiever in all the Mathematics
courses in
the BEngTech Programme
in 2014.
In February 2015, he received the
duties of a PASS Leader (Peer Assisted Study Scheme) after being
nominated by the Head of Engineering School, which qualified
him as
an academic assistant to students that require assistance and tuition
in
their
engineering courses. In terms of his industrial experience, he
completed internships at
the Electricity and Water Authority (EWA) as
a mechanical trainee in 2014. In 2015, Mohammed was part of the
leading upstream petroleum company (Tatweer Petroleum) as a summer
trainee which exposed him to an actual working environment.
Conor
Sheedy
was born in Dublin, Ireland in
1979. He studied at University College
Dublin from 1998 to 2002 receiving a first
class honours Bachelors degree in Electronic
Engineering.
He worked in the Lake
Datacoms department of Lake
Communications from 2002 to 2006, where
he worked as a Digital Signal Processing
Design Engineer specialising in the
development and implementation of Digital
Signal Processing algorithms for embedded
communication systems.
From 2007 to 2011 he worked for the Sino
British College in Shanghai China developing and delivering many of
the subjects for their Electronic Engineering Degree program. In 2010
he received a first class honours Master of Engineering in Electronic
Systems degree from Dublin City University.
He is currently working at
the Bahrain Polytechnic where he has developed and delivered many
subjects on the BEngTech degree.
His research interests include many
aspects of digital communication systems and digital signal processing.
André Hajjar
is the Business Development
Manager for National Instruments in Saudi
Arabia and Bahrain with focus on
government, industrial, and academic
research applications. André graduated with
an Electrical Engineering diploma from the
Lebanese University in Beirut majoring in
Industrial Computing and Electronics. He has
more than 9 years of industry experience and
business development in Medium to High-
Tech businesses. His Business expertise spans
marketing and sales, operations, product conception and R&D. His
Technical expertise spans design, development and commissioning of
system solutions in the areas of Instrumentation, Control and Testing.
Industries covered include Academia & Research, Telecommunications,
Energy & Power, Industrial Control & Instrumentation, Oil and Gas,
Defense and Robotics.
Markets covered are Saudi Arabia, Bahrain and
Lebanon. Development teams managed were located in the USA, India,
Armenia and Lebanon.
339©2016 Journal of Automation and Control Engineering
Journal of Automation and Control Engineering Vol. 4, No. 5, October 2016
Các file đính kèm theo tài liệu này:
- environmental_control_of_a_greenhouse_system_using_ni_embedd.pdf