Sei sulla pagina 1di 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315915067

Comparison methods of Fuzzy Logic Control and Feed Forward Neural Network
in automatic operating temperature and humidity control system (Oyster
Mushroom Farm House) using microcon...

Conference Paper · November 2016


DOI: 10.1109/ISESD.2016.7886713

CITATIONS READS

18 787

9 authors, including:

R. Y. Adhitya S. Kautsar
Politeknik Perkapalan Negeri Surabaya Politeknik Negeri Jember
14 PUBLICATIONS   59 CITATIONS    4 PUBLICATIONS   52 CITATIONS   

SEE PROFILE SEE PROFILE

Noorman Rinanto Ii Munadhif


Shipbuilding Institute of Polytechnic Surabaya 4 PUBLICATIONS   52 CITATIONS   
15 PUBLICATIONS   57 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Developing Safety Integrity Level (SIL) Study for Pertamina Refinery View project

Power Quality Analysis of Integration Photovoltaic Generator to Three Phase Grid under Variable Solar Irradiance Level View project

All content following this page was uploaded by Noorman Rinanto on 29 October 2017.

The user has requested enhancement of the downloaded file.


Comparison Methods of Fuzzy Logic Control and Feed Forward Neural
Network In Automatic Operating Temperature and Humidity Control
System (Oyster Mushroom Farm House) Using Microcontroller
R. Y. Adhitya1), M. A. Ramadhan 1), S. Kautsar1), N. Rinanto1), S. T. Sarena1), Ii Munadhif 1), Mat Syai’in1), R.T.
Soelistijono1), Adi Soeprijanto2)
1) Automation Engineering Study Program, Department of Marine Electrical Engineering,
Shipbuilding Institute of Polytechnic Surabaya (SHIPS/PPNS), Surabaya, Indonesia
2) Department of Electrical Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, Indonesia
Email: ryanyudhaadhitya@gmail.com

Abstract-- The productivity of oyster mushroom of oyster mushroom needs are increased, prompted many
cultivation in low-lying areas are still not optimal. This is actors from UMKM (Micro, Small and Medium home
due to the cultivation of oyster mushrooms needs ideal industry) raise their efforts in fulfill the market demand [1].
temperature and humidity (temperature 22 -28 ° C with a But in reality, the productivity of mushroom cultivation is still
humidity of 60% -80%), while nowadays temperature and considered suboptimal. This is due to the cultivation of
humidity preservation process is done in a conventional mushrooms require ideal temperature and humidity
manner. Given these problems, the researchers gave the (temperature 22 - 28 ° C with a humidity of 60% - 80%) in
solution by creating a tool that able to work automatically order to obtain optimal results. While for now the process of
to monitor and control the temperature and humidity in controlling the temperature and humidity to reach the ideal
oyster mushroom cultivation problem based on mushroom cultivation is done by conventional manner using a
microcontroller. Inputs used in these system are the value manual sprayer and make it more difficult for farmers to
of temperature and humidity data readings from DHT11. obtain the optimal results and when at the dry season it takes
While the output of the system is two actuators, the first is extra energy to secure the conditions of temperature and
the exhaust fan and the second is mist maker. In the humidity of mushrooms house [3].
operation of the appliance automatically there will be two Based on this background, the manufacture of automated
choices of data processing methods are applied, the method tools to control temperature and humidity oyster mushroom
of Fuzzy Logic Control (FLC) and Feed Forward farm house is proposed. Using two data processing methods to
Backpropagation Neural Network (BPNN). Performance find the most optimal and effective method in order to get
tools based on the application of these two methods will be maximum performance result, its controlled by
compared to determine the most optimal and effective microcontroller arduino Mega. This tool is created to give
method when it applied to the tool to automatically control solution for increasing the productivity of oyster mushroom
temperature and humidity oyster mushroom farm house. cultivation in order to remain optimal throughout the year, not
Based on the test results and data analysis, the tools can limited to season, and also the location of cultivation can be
work well and also perform that optimal and effective data carried out in low-lying areas and upland areas.
processing method is Neural Network with an average
conditioning response time of 69.8 seconds to reach the II. METHODOLOGY
ideal temperature and 113.4 seconds for the ideal A. Design System
humidity.
System tools work automatically to control temperature
Keywords: Cultivation of Oyster Mushroom, Fuzzy Logic, and humidity of Oyster Mushroom Farm House, by controlling
Feed Forward Neural Network, Microcontroller, DHT11. environmental conditions such as temperature and humidity to
optimize the growth of oyster mushrooms. Changes in
I. INTRODUCTION temperature and humidity of oyster mushrooms house detected
by the DHT11 sensor. DHT11 sensor data will be used as
Indonesia as an equatorial country, became a heaven for inputs to be processed by the brain of the system, using the
a wide variety of plant kingdom that cause biodiversity in this Arduino Mega with fuzzy logic controller calculation or Feed
country is very abundant. In order to meet national food Forward Neural Network calculation to serve as a reference
security, people should try to develop new sources of food value in taking action in carrying actuators (Exhaust Fan and
types. The mushroom, which was once a wild plant, has now Mist Maker). Data temperature and humidity mushrooms
become a high source of nutrients for humans [2]. One of the house will be displayed on the LCD in real time. For more
main mushroom are much in demand by the public is the details can be seen in the overall system flowchart in Fig. 1.
oyster mushroom (Pleurotus ostreatus). Nowadays the growth

1
Fig. 3. Hardware System

C. Data Processing Method


In this research will explain the application of zero order
sugeno fuzzy logic controller and feed forward neural
Fig. 1. Flowchart of Main System
network. The automatic tools that have been implemented both
B. Hardware Design of these methods will be compared in order to know the most
optimal method to be applied in case of oyster mushroom
The Placement of electronics hardware (such as sensors, cultivation. Input and output parameter values obtained from
controllers and actuators) are arranged in a proportional the results of field studies in one of the main oyster mushroom
condition to get a good response and also more accurate. Here farmers in urban areas, in the Lidah Kulon area, Surabaya. In
is the hardware design of Oyster Mushroom Farm House: addition, according to the literature on the Raharjo’s book with
the title “Oyster Mushroom Cultivation” (Pleourotus ostreatus
var florida) in South Sumatra that takes optimal growth of
mushroom in house temperature is between 28 till 30 ° C and
humidity of 50 till 60% at the time of incubation. While the
temperature in the formation of growing bodies until the
harvest ranged between 22 - 28 ° C with a humidity of 60 -
80%. When the humidity is less, then the plant substrate will
dry. Based on the results of field studies and literatures then
input and output FLC rule base and NN data training
parameters are formed.

Fig. 2. Design of Oyster Mushrooms House D. Application of Fuzzy Logic Method

Oyster mushrooms house design in Figure 2. was designed in The fuzzy logic method by using zero order sugeno
accordance with actual conditions. Roofs made from woven fuzzy logic is applied on the temperature and humidity control
rice straw and walls made from woven bamboo, these devices, the two inputs values, temperature and humidity
materials are used for conditioning air circulation in the values taken from the DHT11. The outputs are two DC
mushrooms house . So the efforts in achieving the set point motors, the first as exhaust fan and the second as Mist Maker.
temperature and humidity can be easily achieved. There are Validating method of data processing is done by using
two input in this system, the temperature and humidity values Matlab. Matlab is used as a calibrator if a micro’s program
obtained from the DHT11. While output in this system there error occurs. Steps that must be done in term to use Matlab as
are two actuators, the exhaust DC fan and mist maker. This is calibrators are:
explained in Figure 3. the hardware system.
1. Making the membership function input.
2. Preparation of the output membership function.
3. Making The Rule.

2
To create the membership function input and output and Figure 7. is a DC fan membership function with three
the rule base is taken from the planning of data processing variables, namely SLOW (5V), MEDIUM (9V) and FAST
methods that have been made previously. (12V). The data generated will be used as a set point DC fan
(exhaust fan).

TABLEI. FLC RULE BASE

Fig. 4. Temperature MF

Figure 4. is a membership function of temperature input


with 5 variables: COLD (15oC - 25oC), COLD ENOUGH
(20oC - 30oC), MEDIUM (25oC - 35oC), HOT ENOUGH
(30oC - 40oC) and HOT (35oC - 45oC). Data to be processed is
the temperature data from DHT11.

Fig. 5. Humidity MF

In Figure 5. is a membership function of humidity input with 5


variables: DRY (0% - 40%), DRY ENOUGH (20% - 50%),
NORMAL (40% - 60%), WET ENOUGH (50% - 60%) and
WET. Data to be processed is the data readout also from the Furthermore, to know about the logic truth of programming
DHT11. performance on Arduino is tested by comparing the output
voltage value of dc fan (exhaust fan) & dc fan (mist maker) on
the serial monitor arduino with dc fan rated output voltage
(exhaust fan) & dc fan (mist maker) in Matlab. Then the error
obtained will be used as a reference for the performance of
fuzzy logic. If the results of testing the application of fuzzy
logic has a small error can be said that Fuzzy has a good
performance, and vice versa if it has a large error, the
performance of fuzzy logic is bad. It can also be proven
through manual calculation by taking one of the main sample
Fig. 6. Mist Maker MF
data readout results from DHT11 sensor readings. Here are the
Figure 6. is a mistmaker membership function with three steps of proving the suitability of the application of fuzzy
variables : SLOW (5V), MEDIUM (9V) and FAST (12V). The methods using manual calculation:
data generated will be used as a set point ultrasonic mist
1. Determine the point of intersection of Triangle Fuzzy
maker.
2. Determine the minimum value. After finding the
intersection point of the triangle fuzzy, the next step is to
compare between temperature’s MF and humidity’s MF,
then determined the minimum value of each MF
comparison.

3. The calculation of the output value by using weight average


method :
Fig. 7. DC fan MF

3
z' =
μ z i i
(1)
TABLEII. NN TRAINING DATA

μ i

E. The Application of Neural Network

Application of Neural Network (NN) has been used


many times to solve any classification or control problems
[4,5,6,7]. The NN temperature and humidity control tools on
oyster mushrooms house have the same input and output
devices like FLC method. Implementation steps Neural
Network method is as follows:

1. Determine the NN method. This paper used Feed Forward


Back Propagation Neural Network as NN method.
2. Determine the maximum error output prediction versus
target.
3. Training process that produces a form Neuron
Architecture.
4. Mean Squared Error analysis (the performance of MSE is
greater than maximum error that has been set before).
5. Take the weight values that are used to implement the
program of microcontroller.

the training process is started by using matlab software to get


the best model based on the training data, figure 9. shows the
mse result is stopped at 24th iteration. It means that the best
training performance is reached at 2,3291 x 10-10 more better
than the maximum error (1 x 10-8) that have been set before.

Fig. 8. Neuron Architecture

Figure 8. is Neuron Arcitecture with 2 inputs derived


from sensor data readings using DHT11,. There are 2 Hidden
layer, each layer have 7 neurons and 2 neurons as output
layer.
Total of primary data used to training section are 25
datas, these data passed the normalization process before it
continue to the training process. Here are the calculation of
normalization method [10]: Fig. 9. Mean Squared Error

The next process can be done by taking each weight in each


v − min layer to implement the calculation of NN on microcontroller.
v' = (new max − new min ) + new min (2)
max− min

4
III. TESTING AND DATA ANALYSIS
Based on the system design that has been explained in
the previous chapter. Here is the hardware design of Oyster
Mushroom house and instrumentation system :

Fig. 12. Panel Box

A. Comparison Methods of Fuzzy Logic and Neural Network


Method
Fig. 10. The Placement of Electronic Components Comparison steps to know the effective application
between Fuzzy Logic and Neural Network in oyster
Fig. 10. shows emergency stop button and running indicator mushroom cultivation problem is done five times with
right on the front cover of panel box. Green light indicates the different temperature and humidity condition, the response
system is running, red light means the system is off. Mist
time is performed from begining condition till required
maker consists of dc fan with modification box and the cold condition 28 ° C and 70% of ideal temperature and humidity.
water inside, if the dc fan is moving then the mist will flow
over the PVC pipelines. LCD is displayed the DHT 11 sensor
value ( temperature and humidity) in real time.

Fig. 11. Exhaust Fan Position


Fig. 13. Temperature Conditioning Performance Comparison Chart
Exhaust fan is located on the front side of oyster mushroom
farm house near the door. It have primary function to stabilize In Figure 13. explain about comparative response time versus
the temperature and humidity inside the farm house by moving begining condition of temperature between Fuzzy Logic and
the hot air inside to the outside. This actutor will stop or keep Neural Network. Feed forward Neural Network clearly more
the latest speed constant which the value of the speed is given superior than FLC at 38oC beginning condition and 36oC in
from the microcontroller in order to reach the setpoint term for conditioning the environment temperature
temperature or humidity.
Based on Figure 14. of the comparative performance
chart of response time versus begining humidity, can be seen
Fig. 12. shows the instrumentation inside the panel box,
that NN more stable in every condition although FLC win the
there are only three components; first 12 VDC power supply
battle in 3rd and 5th condition. But as a whole, the average
for exhaust fan and mist maker, second is 5 VDC power suppy
respone time needs, to reach the set point is dominated by NN
for microcontroller and ARDUINO MEGA 256 as the brain of
at 113.4s in term of humidity and also 69.8s to reach the ideal
the system.
temperature.

5
[5] Mat Syai’in and Adi Soeprijanto, “Neural Network Optimal Power
Flow (NN-OPF) based on IPSO with Developed Load Cluster Method”.
WASET 2010.
[6] Mat Syai’in, Adi Soeprijanto and Eko Mulyanto Yuniarno, “New
algorithm for neural network optimal power flow (NN-OPF) including
generator capability curve constraint and statistic-fuzzy load clustering”
International Journal of Computer Applications 2011. 36(7): p. 1-8.
[7] G. Zhang, B.E. Pattuwo, and M.Y. Hu, “Forecasting with artificial
neural networks: the state of art” International Journal of Forecasting,
vol.14, pp. 35-62, 1998.
[8] Z.L. Sun, T.M. Choi, K.F. Au, and Y.Yu, “Sales forecasting using
extreme learning machine with application in fashion retailing”
Decision Support System, vol. 46, pp. 411-419, 2008.
[9] Chunxia Wang, Chenglin Wen and Yang Lu, “A fault diagnosis method
by using extreme learning machine” IEEE 2015 International
Conference on Estimation, Detection and Information Fusion (ICEDIF
2015).
[10] Paulin, F. And Santhakumaran, A. “Statistical normalization and
backpropagation for classification” International Journal of computer
Fig. 14. Humidity Conditioning Performance Comparison Chart theory and engineering. III(1)

IV. CONCLUSION
TABLEIII. PERFORMANCE COMPARISON TOOL

In this paper has been inducted a Comparison of


Implementation Methods between zero order sugeno Fuzzy
Logic Controller and Feed Forward Neural Network in
operation to control the temperature and humidity of Oyster
Mushroom Cultivation problem in Lowlands Based on
Microcontroller. It can be seen from the results of the testing
tool that compares the response time of the application of
Fuzzy Logic and Neural Network application method with
reference to the value of the initial temperature and humidity.
The result performs that NN is more optimal and effective
with The average response time for conditioning the ideal
temperature and ideal humidity, 69.8 seconds and 113.4
seconds.

REFERENCES

[1] Cahyana YA. Muchordji, M. Bakrun. 2001. Pembibitan,


Pembudidayaan, analisa Usaha Jamur Tiram. Penebar Swadaya.
Jakarta.
[2] Direktorat Budidaya Tanaman Sayuran dan Biofarmaka. Jamur Tiram.
Direktorat Jenderal Bina Jenderal Hortikultura. Jakarta.
[3] Raharjo, Budi (2010). Budidaya Jamur Tiram (Pleourotus ostreatus
var florida) yang ramah lingkungan. BPTP. Sumatera Selatan.
[4] Mat Syai’in, M.F. Adiatmoko, Isa Rachman, L. Subiyanto, “Smart-
Meter based on current transient signal signature and constructive
backpropagation method” IEEE 1st International Conference on
Information Technology, Computer and Electrical Engineering. 2014.

View publication stats

Potrebbero piacerti anche