Potential Fire Hazard Detection System Based on Microcontroller

Using the Fuzzy Mamdani Method

Arif Budi Sampurna

1

, Edy Setiawan

1

, Imam Sutrisno

1

, Urip Mudjiono

1

, Boedi Herijono

1

,

Tri Mulyatno Budhi Hartanto

2

and Ignatius Kristianto Agung Nugroho

3

1

Politeknik Perkapalan Negeri Surabaya, Indonesia

2

Balai Pendidikan dan Pelatihan Transportasi Laut, Indonesia

3

Sekolah Tinggi Ilmu Pelayaran, Indonesia

Keywords: Fuzzy Mamdani, Arduino UNO, MQ-9, MQ-5, MQ-2, AMG8833.

Abstract: Early fire prevention is very important. Based on statistical data from the DKI Jakarta Provincial Fire and

Rescue Service. In 2016 there were 1,047 fire cases, while in 2017 it almost doubled to 2,055 cases. The

danger of a room is also influenced by the level of a gas (based on LEL and UEL values). Therefore, it is

necessary to have a tool to prevent or warn of potential fire hazards early. This final project research makes

potential fire hazard detection system. The sensors used include MQ-9, MQ-5, MQ-2, and thermal sensor

AMG8833. The working system of this tool is AMG8833 connect raspberry to detects fire, output from

AMG8833 will be sent to Arduino as input for Fuzzy Mamdani along with the values from the sensors MQ-

9, MQ-5, MQ- 2 (there are 4 inputs). The result or decision of Fuzzy is the level of potential fire hazard that

occurs. The results of the research are expected to be able to provide early warning or prevention of fire. In

addition, it is hoped that this research module can be used as a teaching module for fire hazard detection

systems to support a better vocational education system. From 10 trials, 90% of the system success was found

for detecting potential fire hazards, for detecting fires using AMG8833 also success and for the success of

sending information via LED-RBG by 90%, and via telegram 100%.

1 INTRODUCTION

Fire is an element that is very useful for humans if

you can control it, but fire can also be dangerous to

humans like fire. Fire is one of the disasters that often

occurs both in big cities and in rural areas that can

cause huge losses (Irawati, 2017). Fires can occur at

any time without anyone being able to predict them.

Many factors can affect the occurrence of fires,

including electrical short circuits, temperature

increases from mechanical processes that cause fires,

and human factors or human error (Wijaya, et al.,

2020).

In a company or housing, it is very important to

have a means of detecting the potential level of fire

hazard. This tool aims to detect the potential level of

fire hazard early on so that it can be overcome quickly

before a fire occurs. Many fire disasters have caused

many casualties and property losses. Fires in homes

often occur due to accidents in the kitchen

environment, or electrical problems in the installation

of electronic devices. Fires start at a small level but

can endanger large houses or surrounding buildings

(Irawati, 2017).

Based on statistical data from the DKI Jakarta

Provincial Fire and Rescue Service, there were 1,047

fire cases in DKI Jakarta province in 2016. A total of

754 cases were caused by electricity, 35 cases due to

cigarettes, 75 cases due to stoves and 183 cases due

to other factors and 0 for unknown cases, but in 2017

fire cases increased almost 2 times to 2,055 cases. A

total of 851 cases were caused by electricity, 33 cases

were caused by cigarettes, 156 cases were caused by

stoves, 1009 cases were caused by other factors, and

6 cases had no known cause. While the results of

statistical data on fire incidents in Indonesia in 2007,

most fire accidents occurred in residential areas,

namely 65.8%, shopping centers 9.8%, industrial

buildings by 8%, offices by 5.6%, markets by 4.8%,

hotels by 4.6 %, and other buildings by 0.4% (Agusri,

2021).

The room can be said to be flammable if a certain

volume of hazardous gas has passed the Lower

Explosive Limit (LEL) or the lower limit of the

300

Sampurna, A., Setiawan, E., Sutrisno, I., Mudjiono, U., Herijono, B., Hartanto, T. and Nugroho, I.

Potential Fire Hazard Detection System Based on Microcontroller Using the Fuzzy Mamdani Method.

DOI: 10.5220/0011762500003575

In Proceedings of the 5th International Conference on Applied Science and Technology on Engineering Science (iCAST-ES 2022), pages 300-306

ISBN: 978-989-758-619-4; ISSN: 2975-8246

explosion and can be said to be very dangerous or

very easy to fire if it has passed the Upper Explosive

Limit (UEL) or the upper limit of explosion.

According to data from CHRISALIS SCIENTIFIC

TECHNOLOGIES INC. the lower explosion limit

(LEL) of methane gas in a room is 5% of the total

volume and the upper limit (UEL) of 15% of the total

volume of the room, the lower limit of the explosion

(LEL) of carbon monoxide gas is 12.5% of the total

volume and above (UEL) 74% of the total volume of

the room. Therefore, a fire hazard detection system is

needed that works properly to prevent fires (Amanda,

2021).

In this study, a system that can detect the potential

level of fire hazard from an early age will be made

using 3 types of parameters using the Fuzzy Mamdani

method (Batubara, 2017). Information on the

potential level of fire hazard will be sent via telegram.

It is hoped that this system can reduce or prevent fires

so that they can reduce the number of fatalities and

minimize losses.

The formulation of the problem in this study

include:

How to detect potential fire hazard level with

Fuzzy Mamdani using AMG8833, MQ-9, MQ-5,

MQ-2, and DHT22 sensors?

How to detect fire using AMG8833 sensor?

How to convey information that there has been a

potential fire hazard?

2 METHOD

The stages carried out in this study are depicted in the

flow chart in Figure 1 below

Figure 1: Research Flowchart.

Based on the flow chart above, the detailed flow

of research activities will be described in the

following sub-chapters

2.1 System Analysis

The following are the hardware and software

requirements used to build the overall system.

Hardware Software

Sensor MQ-9

Sensor MQ-5

Sensor MQ-2

Sensor AMG8833

Raspberry Pi 4B

Arduino Uno

Buzzer SFM27

LED RGB

Water Pump

Smart

p

hone

Arduino IDE

Python

MATLAB

Eagle

This system has four input parameters. In the

input or input section there are 4 sensors, namely the

MQ-9, MQ-5, MQ-2 sensors to detect dangerous

gases and the AMG8833 sensor to detect fire, the four

sensors are integrated using the Fuzzy Mamdani

method on the Arduino Uno to detect whether there is

a potential danger fire or not. (Febriany, 2016). The

results of the system will determine the potential fire

hazard level between levels 0 to 4, if the system states

a fire hazard potential level 0 then nothing happens,

while if the system states a fire hazard potential level

1 or 2 or 3 or 4 the LED will light up (green for level

1, yellow for level 2, orange/orange for level 3 and

red for level 4), the water pump will pour water (if

fire is detected) and the raspberry will send a

message/notification to the smartphone (telegram) .

Then when the level of one or more harmful gases or

exceeds the safe limit, the system will sound a buzzer.

The concept of the system is illustrated in the block

diagram in Figure 2 (Mahendra, 2021).

Figure 2: System Concept.

Potential Fire Hazard Detection System Based on Microcontroller Using the Fuzzy Mamdani Method

301

Figure 3: Mechanical Design Results.

The workflow of the system in this study can be seen

in Figure 4 below

Figure 4: System Workflow Diagram.

2.2 Fuzzy Logic

The application of fuzzy logic to the system is used to

identify the potential level of fire hazard. In this

study, the fuzzy mamdani method was used with 4

inputs and 1 output in the form of identification.

Fuzzy in this system will be processed by a

microcontroller with input in the form of carbon

monoxide gas from the MQ-9 sensor reading,

methane gas from the MQ-5 sensor reading, propane

gas reading from the MQ-2 sensor and the fire

temperature reading from the AMG8833 sensor.

While the output is in the form of identifying the level

of potential fire hazard from level 0 to level 4

(Muhathir, 2021).

3 RESULTS AND DISCUSSION

In this design, there are inputs in the form of carbon

monoxide gas from the MQ-9 sensor readings,

methane gas from the MQ-5 sensor readings, propane

gas readings from the MQ-2 sensor and fire

temperature readings from the AMG8833 sensor.

Then the output is in the form of identification of the

potential level of fire hazard from level 0 to level 4.

Here is the membership function of Fuzzy logic input

and its output.

Figure 5: Membership Function Gas CO.

Figure 6: Membership Function Gas CH4.

Figure 7: Membership Function Gas C3H8.

Figure 8: Membership Function Fire Temperature.

Figure 9: Fuzzy Logic Output.

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

302

3.1 Fuzzy Logic Test

In modeling the Fuzzy logic method that will be used

for the control process of the system in this final

project, a simulation is needed to determine the input,

output, or rules that will be used. Matlab software is

used for fuzzy logic simulations either with toolbox,

m-file, or simulink (Suyono, 2017).

Fuzzy testing is done by comparing the fuzzy

output with the results of matlab and the results of

manual calculations (Sutrisno, 2014).

Table 1: Fuzzy Logic Inference.

CO CH4 C3H8 Fire

Danger

Level

A A A None 0

A A A Small 1

A A A Big 2

A A B None 1

A A B Small 2

A A B Big 3

A A SB None 2

A A SB Small 4

A A SB Big 4

A B A None 1

A B A Small 2

A B A Big 3

A B B None 1

A B B Small 2

A B B Big 3

A B SB None 2

A B SB Small 4

A B SB Big 4

A SB A None 2

A SB A Small 4

A SB A Big 4

A SB B None 2

A SB B Small 4

A SB B Big 4

A SB SB None 3

A SB SB Small 4

A SB SB Big 1

B A A None 2

B A A Small 3

B A A Big 1

B A B None 2

B A B Small 3

B A B Big 2

B A SB None 4

B A SB Small 4

B A SB Big 1

B B A None 2

B B A Small 3

B B A Big 1

B B B None 2

B B B Small 3

B B B Big 4

B B SB None 3

B B SB Small 4

B B SB Big 4

B SB A None 2

B SB A Small 4

B SB A Big 4

B SB B None 3

B SB B Small 4

B SB B Big 4

B SB SB None 3

B SB SB Small 4

Potential Fire Hazard Detection System Based on Microcontroller Using the Fuzzy Mamdani Method

303

Table 1: Fuzzy Logic Inference (cont.).

CO CH4 C3H8 Fire

Danger

Level

B SB SB Big 4

SB A A None 2

SB A A Small 4

SB A A Big 4

SB A B None 2

SB A B Small 4

SB A B Big 4

SB A SB None 3

SB A SB Small 4

SB A SB Big 4

SB B A None 2

SB B A Small 4

SB B A Big 4

SB B B None 3

SB B B Small 4

SB B B Big 4

SB B SB None 3

SB B SB Small 4

SB B SB Big 4

SB SB A None 3

SB SB A Small 4

SB SB A Big 4

SB SB B None 3

SB SB B Small 4

SB SB B Big 4

SB SB SB None 4

SB SB SB Small 2

SB SB SB Big 4

3.2 Telegram Test

In this test, to find out whether the telegram message

that was sent was successfully sent by the system.

Figure 10: Telegram Test.

3.3 Water Pump Test

Table 2: Testing Water Pump.

Test Water Pump

There's a fire turn on

In Table 2. shows the results of the Water Pump test

to pour water when a fire is detected. The results of

testing the condition that there is fire, it was found

that 10 attempts were made with 10 successes and 0

failures. The resulting error is 0% (Sutrisno,.2012).

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

304

3.4 Buzzer Test

Table 3: Buzzer Testing on Gas Levels.

Test Buzze

r

There is a dan

g

erous

g

as level turn on

There is a dan

g

erous

g

as level turn on

There is a dangerous gas level turn on

There is a dan

g

erous

g

as level turn on

There is a dan

g

erous

g

as level turn on

There is a dan

g

erous

g

as level turn on

There is a dan

g

erous

g

as level turn on

There is a dangerous gas level turn on

In Table 3. shows the results of the Buzzer test to give

a warning when gas levels are detected that exceed

the safe limit. The results of the condition test have

gas levels exceeding the safe limit, 10 trials were

found with 10 successes and 0 failures. The resulting

error is 0% (Jami'in 2014).

3.5 LED-RGB Test

LED-RGB is used as an indicator of the output of the

fuzzy mamdani method.

Table 4: LED-RGB Test.

Test LED-RGB

Hazar

d

Potential Level 0 No flame

Hazar

d

Potential Level 0 No flame

Hazar

d

Potential Level 1 Green Light

Hazar

d

Potential Level 1 Green Li

g

ht

Hazar

d

Potential Level 1 Green Li

g

ht

Hazar

d

Potential Level 2 Yellow Li

g

ht

Hazar

d

Potential Level 2 Yellow Light

Hazar

d

Potential Level 3 Re

d

Light

Hazar

d

Potential Level 4 Re

d

Light

Hazar

d

Potential Level 4 Re

d

Light

In Table 4. shows the results of the Buzzer test to give

a warning when gas levels are detected that exceed

the safe limit. The results of the condition test showed

that there was a potential fire hazard, 10 trials were

found with 9 successes and 1 failure. The resulting

error is 10%.

3.6 Overall System Test

In testing the overall system, a room is carried out by

giving fire and gas using a gas lighter.

Table 5: Overall System Test.

CO CH4 C3H8 Temp Lv LED Pump Buzz

19.08 1.15 6.60 28.25 0 Off M M

19.88 2.59 2739.7 28 2 K M N

1344.8 198.67 36.03 28.0 2 K M N

34.18 181.91 778.46 28.0 1 H M N

18.72 4.94 12.19 34.0 1 H N M

18.37 3.81 10.69 122.0 2 K N M

26.29 715.25 82.33 80.5 3 O N N

30.56 1254.5 51.58 86.0 4 Off N N

400.15 687.54 754.12 29.0 1 H M N

1694.1 50.17 954.87 28.25 3 O M N

In Table 5 shows the results of testing the entire

system. Output on the LED, K is yellow, H is green,

O is orange, M is red. Output at pump and buzzer, M

is off, N is on. It was found from the results of 10

tests, there was 1 incorrect result or a 90% success

rate (Sutrisno, 2014).

4 CONCLUSION

Based on the results of the tests that have been carried

out in this study, it can be concluded that:

The application of Fuzzy Mamdani on a system

with four input parameters including carbon

monoxide gas, methane gas, propane gas and fire

temperature can detect potential fire hazard levels

with an accuracy of 90%.

The application of the AMG8833 sensor to detect

fire, the AMG8833 sensor reads the heat value or

temperature of each pixel and then takes the hottest

value as the fire temperature input.

The application of delivering information on

potential fire hazards via LED-RGB has an accuracy

Potential Fire Hazard Detection System Based on Microcontroller Using the Fuzzy Mamdani Method

305

of 90% and the delivery of information via telegram

is sent once every 10 seconds with 100% accuracy.

SIM800l Communication Testing to databases and

websites does not occur sending data errors or

without errors with a delay of 10 seconds so there is

no you can say real time.

The suggestions that can be given from this

research include:

The use of thermal sensors to detect fire is good

enough, but it would be better if there is a camera

sensor that can detect fire in real.

Testing the sensor would be better to use a gas that

really matches the gas being detected.

REFERENCES

Agusri, E., Hasan, M. N. R. & Martini, R. S., 2021. Analisa

Kebutuhan Air Bersih Untuk Operasional Harian Dan

Sistem Pemadam Kebakaran Sprinkler Gedung Utama

Baru Rumah Sakit Bhayangkara Palembang. Volume

10, pp. 338-349.

Amanda, N., 2021. Rancang Bangun Alat Pendeteksi Titik

Api Dan Asap Menggunakan Metode Yolo Dan

Berbasis Android Untuk Deteksi Dini Kebakaran Pada

Gedung Bertingkat. Surabaya: Politeknik Perkapalan

Negeri Suarabaya.

Batubara, S., 2017. Analisis Perbandingan Metode Fuzzy

Mamdani Dan Fuzzy Sugeno Untuk Penentuan

Kualitas Cor Beton Instan. Medan: Universitas

Pembangunan Panca Budi Medan.

Febriany, N., 2016. Aplikasi Metode Fuzzy Mamdani

Dalam Penentuan Status Gizi Kebutuhan Kalori Harian

Balita Menggunakan Software Matlab. pp. 29-49.

Irawati, R., 2017. Model Peringatan Kebakaran Dengan

Fuzzy Mamdani. pp. 30-37.

Mahendra, A., 2021. Prototype Deteksi Kebakaran

Menggunakan Kamera Dengan Notifikasi Message

Telegram Berbasis Iot Di Bandara Mutiara Sis Al-

Jufrie. Surabaya: Politeknik Penerbangan Surabaya.

Muhathir, Santoso, M. h. & Lestari, D. A., 2021. Wayang

Image Classification Using SVM Method and GLCM

Feature Extraction. JITE (Journal of Informatics and

Telecommunication Engineering), pp. 373-382.

Suyono, H., Setyawati, O. & Dirgantara, W., 2017. Sistem

Peringatan Dini untuk Deteksi Kebakaran pada

Kebocoran Gas Menggunakan Fuzzy Logic Control.

pp. 27-32.

Wijaya, I. D., Anam, S. & Rismanto, R., 2020. Rancang

Bangun Sistem Deteksi Dan Pemadam Kebakaran Pada

Smart Home Menggunakan Metode Fuzzy. pp. 9-16.

Imam Sutrisno, Mohammad Abu Jami’in, Jinglu Hu.

(2014). Modified fuzzy adaptive controller applied to

nonlinear systems modeled under quasi-ARX neural

network, Artificial Life and Robotics, vol. 19, issue 1,

pp. 22-26.

Sutrisno, Imam. (2012). Neural predictive controller of

nonlinear systems based on quasi-ARX neural network,

ICAC 12 - 18th International Conference on

Automation and Computing.

Jami'in M.A., Sutrisno I. (2014). An adaptive predictive

control based on a quasi-ARX neural network model.

13th International Conference on Control Automation

Robotics and Vision, ICARCV

Sutrisno, Imam. (2014). An improved adaptive switching

control based on quasi-ARX neural network for

nonlinear systems. Artificial Life and Robotics.

iCAST-ES 2022 - International Conference on Applied Science and Technology on Engineering Science

306