Sensors network in the greenhouse
Witold Kopacz
Wrocław University of Technology
Wyb. Wyspiańskiego 27
50-370 Wrocław, Poland
Students Scientific Association TRAF
wkopacz@wp.pl
Bartosz Kołodziej
Wrocław University of Technology
Wyb. Wyspiańskiego 27
50-370 Wrocław, Poland
University of Almería
Faculty of Computer Computation
La Cañada de San Urbano
04120 Almería, Spain
kolodziej.wroc@gmail.com
Abstract
The paper presents an automatic acquisition sensor
network which acquires environmental information
about weather conditions in the greenhouse. The
network is built on the basis of devices produced by the
National Instruments company. The system was
designed at the University of Almeria (Spain) and
installed in Las Palmerillas research center. It allows
to remote surveillance of the main environmental
conditions such as temperature, humidity, global
radiation
and
PAR
(Photosynthetically
Active
Radiation), the concentration of CO
2
in the building,
as well as the transmission of information to the data
storage device. It controls the production process by
taking charge of the yield. The system enables the
acquisition of data necessary for the proper regulation
of control devices, which were designed to regulate the
internal conditions in the greenhouse according to the
regime of production. The information obtained from
the sensors allows the conditions inside the greenhouse
to maintain independent of external factors.
1. Introduction
There are many well known systems which acquire
data from sensors in the greenhouses agriculture. A
large part of them is based on the solutions invented
and developed specifically for this type of objects. The
solutions adopted in greenhouses are characterized by
their resistance to specific environmental conditions,
such as high temperature or humidity and strong solar
radiation. Considerable number of types of network
topologies and standards induces problems in the
transfer of solutions between different objects.
An important prerequisite for success in the
cultivation of greenhouse plants, is appropriate control
of acting devices which are installed in a greenhouse.
These include ventilation flaps, solar curtains, gas
burners, heating and equipment for the artificial
enrichment of CO
2
in the atmosphere of greenhouses.
Typically the PI or PID controller are responsible for
control of those devices [1-3]. The standard activity
for these types of controls is to minimize the error in
the steady state. The error in steady state, is the
difference between the current value and a given value
of some environmental condition. The desired value of
controlled variable results from the regime of
production and it is based on experiments carried out in
multi-center study of plant cultivation. To know the
value of the current environmental condition, it is
necessary to use sensors to measure the value of the
examined parameter in suitable units, and then in the
form of electrical signal (most often) send it to the
controller and data storage unite.
To transmit the
signals from sensors to controllers and data storage
devices, it is necessary to use an efficient transmission
network.
Our aim is to provide possible solutions to this type
of sensor networks. This project was executed in the
research center Las Palmerillas [4], where it has been
working since April 2008, It has already presented the
effectiveness on satisfactory level for its intended
application.
2. Type of acquired data
As already been mentioned, the task of sensor
network in a structure, is to obtain information about
the
current
environmental
conditions
in
the
greenhouse. The possibility of application an adaptive
PI or PID controller impose the necessity to collect
additional
information
on
weather
parameters,
monitoring the external conditions that have a major
impact on the value of the controller settings in case of
adaptive control [5]. The information necessary to
determine the state of the environment inside the
building include: air temperature, air humidity and the
concentration of CO
2
. In addition, to ensure
appropriate conditions for growing plants, it is also
essential to collect information about the global solar
radiation and PAR radiation, which has the primary
influence on the process of photosynthesis in plants.
Selection of the settings of PI and PID controller in
the process of adaptation to external conditions,
depends in a large extent on the direction and speed of
wind outside the greenhouse [1,2]. Bearing in mind
these two groups of information about the factors that
affect the facility, a network of sensors based on both
the sensors inside the greenhouse as well as the
external weather station was projected and finally built.
3. Types of selected sensors
In framework the control system implementation is
based on two level structure [5,6,7]. This is a kind of
hierarchical structure, each task is handled with a
specific control. This kind of structure has been tested
in other types of solar powered systems, like in the
PSA
(Plataforma
Solar
de
Almeria).
This
decomposition can provide many benefits, such as
production costs optimization, or reducing the
possibility of human error. This multilayer control
structure has been indicated on Figure 1.
Figure 1. Multilayer control structure.
The upper layer is based on the optimal production
schedule. It adjusts the desired environmental factors
to the season and to the type of produced crop. The
lower regulatory control layer is related to the control
of internal temperature an humidity. It gets the desired
internal conditions provided by the upper layer (the
Reference Governor Unit). Using the PI control, it
actuates with the ventilation, heating and blinds
correcting the transient disturbances. To get the
information about the internal conditions, it relays on
the data provided by the data acquisition system.
The data acquisition system consists of two basic
levels: sensor level, and the data transmission network
level. Another part of the system is the controller and
the data storage device based on the PC computer,
which in the production schedule level. The
construction of the system has been indicated in
Figure 2.
Figure 2. Acquisition system structure.
In the sensor part thermometers combined with
humidity sensors can be highlighted. They act on the
basis of the combined EE21-FT6A elements. This level
also includes devices for measuring the global-type
radiation MRG-1P produced by ITC, the devices for
measuring radiation PAR-type LI-190 produced by
Li-Cor, and the CO
2
concentration sensor AU-06
produced by PRIVA. Acquisition of data on wind speed
and direction is supported by the 03002 sensor,
produced by the company Young. Signal output for the
above sensors are electrical values, voltage signal in
range of 0-10V for the measurement of humidity,
temperature, wind direction, global radiation and PAR
radiation, and the current signal in the standard range
of 4-20mA for the measurement of the concentration of
CO
2
. The measurement of wind speed shall be made by
counting pulses generated by the anemometer in one
second. Each sensor is powered by a unique source,
which separates the device from the electricity
network. The sensors were placed in the central part of
the greenhouse. This position was chosen after
modeling the temperature distribution inside the object.
It provides a minimal effect on the temporary outdoor
conditions which might affect the measurements. Such
positioning of the sensors will also consider the impact
of growing plants on the distribution of environmental
parameters in the object. Output signals from the
sensors are supplied by cable to input devices
(cFP modules), processing the analog signal to digital
signal, and then forming the data to the standard
frames which can be send using the Local Area
Network (LAN).
4. Interfacing devices
In the interfacing devices, the data collected from the
sensors are transformed into a form which will be
readable to the controlling program of the greenhouse.
The idea was to design a wide-use equipment available
on the market for industrial type PLC. It has been
decided to use a modular data acquisition system type
CompactFieldPoint
(cFP)
[8],
produced
by
National Instruments. The advantage of this system is
great potential for expansion and modification, as well
as the possibility of cooperation with the LabView [8]
program, in which the program controlling the
greenhouse was written. The whole set consists of a
network interface modules and input/output extension
modules. The module is attached to a network Ethernet
cable, fitted with a standard RJ45 plug. The voltage or
current signal leads to the I/O module through the
screw connectors slot. To launch the system is
necessary to configure the module, and network
cards. This can be done by using the Measurement &
Automation Studio [8], which is delivered by the
manufacturer, along with components of the system.
Since the sensor network was built in two greenhouses,
it has become necessary to apply the network switch.
Each greenhouse has a separate switch, but only this,
which is located in a greenhouse NAVE7 is connected
directly to a PC, which controls the cultivation. The
physical construction a diagram of sensor networks is
shown in Figures 3 and 4.
Figure 3. Structure of the sensors network in NAVE6.
Figure 4. Structure of the sensors network in NAVE7.
Elements marked on the drawings by the numbers "1",
"2" and "3" are the CompactFieldPoint network
modules, attached to the I/O cards. The module
number 1 collects the external measurements from the
weather stations(WS): wind direction(A), wind
speed(W),
outdoor
temperature(T),
humidity(H),
global radiation(Glob) and PAR radiation (PAR ). To
the network interface(cFP1804) are attached extension
modules cFP AI-100 and cFP DI-300. The module
number 2 collects the data describing the environment
inside a greenhouse NAVE7: the concentration of
CO
2
(CO2), temperature(T), humidity of the air
inside(H),
global
radiation(Glob)
and
PAR
radiation(PAR). The network interface(cFP1804) is
supplemented by the input card cFP AI-112. Module
number 3 provides similar information on the
environment inside the greenhouse NAVE6, and it is
also connected to the cFP card AI-112. During the tests
verifying the accuracy of the design and assembly,
some problems related to the data transportation in the
LAN network between the PC controller and the
various network modules have been found. Thus it was
decided to use an redundant network switch in the
greenhouse NAVE6, which should improve the quality
of the signal in the above-mentioned network.
System architecture is based on assigning to each
network interface (CompactFieldPoint 1804) an unique
IP address, which will be identified in the LAN. In
addition, each I/O module is assigned a unique
identifier that allows to cooperate with the controller
written in LabView environment. In the greenhouse
NAVE6 one network interface module has been
adapted, and in greenhouses NAVE7 two modules, due
to the need to collect information from an external
weather station. Aggregated elements of the system are
shown in Table 1.
Table 1. List of data acquisition modules.
Type
Quantity
Description
cFP1804
3
LAN interface
cFP AI-100
1
Analog input
module
cFP AI-112
2
Analog input
module
cFP DI-300
1
Digital input
module
Due to difficult working conditions and the possibility
of flooding by chemicals, it was necessary to bring all
the elements of the system together in airtight PVC
boxes, with the front wall made of transparent material,
to allow observation of the state of the equipment.
Module cFP1804 allows to connect to the Ethernet
standard 10Mb/s or 100Mb/s up to four extension cards
from the cFP family. The module has an RJ45 type
jack to connect it to a standard 10BaseT or 10BaseTX
network. During the transmission , the TCP/IP protocol
is used.
5. Tests and experiments
After installation of the equipment in the object,
tests to assess the operation of the sensors have been
launched. As an experiment to test a network of
sensors, a standard method of identifying the dynamics
of the system involving the introduction on the entry of
the system a boost in the form of stroke, and then
analyze the system response. The impulse in the
present case was the variable position of the ventilation
flaps, which should result in changes of air temperature
and humidity inside the greenhouse, with the speed
depending on the dynamics of the object. Changes in
these parameters should be measured by sensors and
then sent through the network to the controller and the
data storage device. Data collected in such way, may
be further examined and analyzed to determine the
parameters of the system. Using the data obtained in
this experiment, it is possible to assess the operation of
sensor networks. In particular, it may useful evaluate
the dynamics of the sensor network, namely the speed
of
their
response
to
changing
environmental
parameters in which they are located.
The experiment
was carried out several times with opening (Figure 5),
and closing (Figure 6) ventilation flaps. The results of
the measurements in the form of graphs were shown in
the figures. The results were compared with the old
measurements. This comparison shows that the data
accuracy is good enough to install the examined
system in new greenhouses.
Figure 5. Temperature while opening the ventilation.
Figure 6. Temperature while closing the ventilation.
During the experiment the outgoing data and results
of the measurements were monitored, as well as the
condition of equipment. Acquisition and storage of
data collected from the sensor networks proceeded
without problems. Results varied in the expected range.
6. Conclusions
The sensor network described in this paper works in
Las Palmerillas experimental greenhouses for over one
year. Experiences and the results of cultivation confirm
its usefulness and reliability. Specific conditions in the
greenhouse did not cause any difficulties. Archive in
Las Palmerillas and the University of Almeria already
have several hundred megabytes of data obtained
through the work of this network. On the basis of these
materials, a mathematical model of greenhouse
environment was developed, as well as appropriate
controller was chosen and gained. Both the model and
the controller are described in work [2,3]. The
appropriate control of internal conditions in the
greenhouse allows the final user to reduce the amount
of fertilizers and pesticides utilize during the
production. The network of sensors and interfacing
devices has a modular design, which implicates the
possibility of further extension and improvement
activities. As it was expected, during the work on
system design a great help and support, both in online
resources as well as in the literature sources could be
found. It should be noticed that communication
between remote sensors in two greenhouses (NAVE6
and NAVE7), sharing some resources such as the
LAN, and the PC controlling the process of cultivation
is still, after one year of continuous operation, very
stable. This is important in the perspective of
automation another similar greenhouse by using a
similar sensor network to the one designed in 2008.
This new investment is planned for June 2009, and the
experience gained during previous work
will
significantly simplifies the process of designing the
system.
7. References
[1] P.G.H. Kamp G. J. Yimmerman, Computerized
Environmental Control In Greenhouses, IPC Plant Ede - The
Nedherlands, 1996.
[2] Francisco Rodriguez Diaz, Modelado y control jerarquico
de crecimiento de cultivos en invernadero(in Spanish),
Escuela Politecnica Superior de la Universidad de Almeria,
2002.
[3] Bartosz Kołodziej, Komputeryzacja szklarni(in Polish),
Politechnika Wrocławska, Wrocław, September 2008.
[4]
http://www.laspalmerillas.cajamar.es/Default.htm
,
La Estación Experimental de la Fundación Cajamar,
May 2009.
[5] A. Pawlowski, J.L. Guzman, F. Rodríguez, M. Berenguel,
J. Sánchez, S. Dormido, “Simulation of Greenhouse Climate
Monitoring and Control with Wireless Sensor Network and
Event-Based Control”, Sensors, Sensors Editorial Office,
Basel, Switzerland, January 2009, pp. 232-252.
[6] R. Klempous, J. Kotowski, J. Nikodem, J. Ułasiewicz,
“Optimization algorithms of operative control in water
distribution systems”, Journal of Computational and Applied
Mathematics 84, Elsevier, pp. 81-99, 1997.
[7] C.M. Cirre, M. Berenguel, L. Valenzuela, R. Klempous,
“Reference governor optimization and control of a
distributed solar collector field”, European Journal of
Operational Research 193, Elsevier, 2009.
[8]
http://www.ni.com/pdf/products/us/2005_6072_161_101_
d.pdf
, National Instruments Corporation, 2006.