Design of Alcohol Detection and Classification Devices in Traditional

Legen / Tuak Drinks using an IoT-based MQ-3 Sensor

Diana Rahmawati

, Koko Joni

, Rahmah Syarifah Febriana

and Heri Setiawan

Electrical Engineering Dept, University of Trunojoyo Madura, Bangkalan, Indonesia

Electrical Weapon System Dept, Polytechnic of Indonesian Army, Batu, Indonesia

Keywords: ATmega2560; Classification of alcohol; IoT ; MQ-3; Naïve Bayes.

Abstract: Rapid detection and classification of alcohol content in traditional legen / tuak drinks need,because high

alcohol content in drinks is very dangerous for consumption. MQ-3 sensors used to obtain electronic aspects

of chemical reactions then captured by the ATmega2560 microcontroller as data retrieval of objects, which

will then be forwarded

for processing on a Personal Computer. This tool is designed to display and classify

legen/palm wine with 3 conditions (good, good

enough, dangerous). The IoT(Internet of Things) technology

is used with a short

process and display data accurately

on smartphone

. Using the naïve Bayes method, the

accuracy of the tool in the trial results is 90%

successful and it can be said that this tool functions well.

1 INTRODUCTION

Nira or roomie is a sweet liquid obtained from the

stems of plants such as sugar cane, beetroot, sorghum,

maple or bunches of sap from the palm family such

as sugar palm, palm, date, sago, siwalan and so on.

Palm palm juice or commonly called "legen" this

word is actually

the term "legi" in Javanese which

means "sweet". In the process of tapping the roomie

needs to be done with good handling and

afterward,

roomie is a liquid containing certain sugars (sucrose,

glucose, fructose, and carbohydrate) which has an

average acidity level

around 6-7 and has a fragrant

aroma. If the roomie is stored in a period of time

there will be natural fermentation by the presence

microorganisms contained in the roomie, thus

producing an acidic taste due to the formation of

acetic sour which is a good medium for

the growth of

microorganisms such as bacteria. In a vulnerable

short storage time the drink when consumed will

have a negative

impact due to the longer storage can

lead to the activity of enzymes that are in the roomie

develops and makes the alcohol content in these

drinks increase. Sunanto says that in Indonesia palm

trees can grow well and be able to produce in areas

with fertile soil at an altitude of 500m-800m above

sea level. In areas that have a height of less than

500m or more than 800m, sugar palm plants can still

grow but the fruit production is less than the

maximum. Tuak is a typical drink that is tapped from

palm trees and then stored for 6 hours to 7 hours so

it undergoes a fermentation process and turns into a

drink that has an alcohol content of 4% -5%. Sweet

Tuak is a drink that contains alcohol and is a type of

traditional drink made from palm rommie (Bhuta,

Desai, & Keni, 2015)(Alkohol & Fermentasi,

n.d.)(Ikegami, 1997)(I. G. Ayu, Dhyanaputri, & Jirna,

2017). Palm trees are also referred to as tuak trees,

producing palm water (sap) that drips from the flower

arrangements. People limit the notion of fermentation

only to alcoholization and lactation (Fatmawati,

2016)(Rizal, Erna, Nurainy, & Tambunan, 2016).

Fermentation is an anaerobic overhaul of

carbohydrates that

results in the formation of stable

fermented products(Rizal et al., 2016) (Ilmu &

Dalam, 2019)(Pamungkas & Kompiang,

2006)(Moede & Gonggo, 2017)(I. Ayu, Pranayanti,

& Sutrisno, 2015) .

The Regulation from the Indonesia’s Minister of

Health No.86 / 1977 say that alcoholic drinks are

divided into 3 groups, namely A with 1-5% alcohol

content, class B with 5-20% alcohol content and C

group with 20-55% alcohol content.(Artikel,

2018)(Pangan & Pelita, 2017) To

find out the alcohol

content in drinks, labolatorium testing requires and

take a long time(Alkohol & Fermentasi, n.d.). In this

case the traditional

drink of sap or "legen" is a drink

that does not contain alcohol, prior to the storage

process, it is uncertain the percentage increase in

alcohol content in this drink.(I. G. Ayu et al., 2017)

278

Rahmawati, D., Joni, K., Febriana, R. and Setiawan, H.

Design of Alcohol Detection and Classiﬁcation Devices in Traditional Legen / Tuak Drinks using an IoT-based MQ-3 Sensor.

DOI: 10.5220/0010331302780284

In Proceedings of the International Conference on Health Informatics, Medical, Biological Engineering, and Pharmaceutical (HIMBEP 2020), pages 278-284

ISBN: 978-989-758-500-5

Not a few ordinary Muslim people who are mistaken

in knowing the amount of alcohol taken from the

fermentation of roomie. From the LPPOM MUI

2008 notes, beverage products from the fermentation

process that contain alcohol

(naturally present) are

allowed to be consumed if the amount is very small at

less than 1%(I. G. Ayu et al., 2017)(Alkohol &

Fermentasi, n.d.).

Figure 1: system design

2 METHOD

2.1 System Design

In order to detect and classify of alcohol content we

use Arduino as microcontroller. For the classification

process use 3 (three) conditions, namely "good",

"good enough", "dangerous". ATmega2560

microcontroller get the power supply from the

battery/power and used as supply. The MQ-3 Sensor,

used to detect ethanol (alcohol) content in the form of

Analog To Digital Converter (ADC) signals(Alkohol

& Fermentasi, n.d.). PH Probe electrode and pH

modules are used to detect and relating tools from

sensor to the micro. DS18B20 serves as a limitation

of the temperature parameters that the sample testing

can perform. NodeMCU used to be a transmitter or

data sender from the Arduino to the server and as the

media viewer that read the ADC voltage output of the

MQ-3 sensors which will be displayed on the LCD as

a percentage of content, temperature magnitudes, pH

content, sample conditions and forecasting samples in

real time.

2.2 Testing

By using the Naïve Bayes method, the learning

methods that use probability calculations, we test this

system. The algorithm utilizes simple probability and

statistical calculations, thinking that inter-class

classes with other classes can stand alone on the other

classes. Naïve Bayes is a method that has no rules,

using a branch of mathematics known for probability

theory to find the greatest opportunity of the

classification possible by looking at the frequency of

each classification on the training data. Naïve Bayes

is a popular classification method and is included in

the top ten algorithms in data mining, the algorithm is

also known as Idiot's Bayes, Simple Bayes an

Independence Bayes (Lubis & Pinem, 2014)(Profile,

2018)(Raschka, 2014) of Bayes ' classification based

on Bayes ' theorem, taken from the name of a

mathematician who was also the British minister of

Prebysterian, Thomas Bayes (1702-1761) (Profile,

2018)Here's the equation of the Naïve Bayes

theorem:

P ( X | Y ) = …………………………… .. (1)

Information:

Y: data with unknown classes

X: hypothesis data y is a specific class

P (x|y): hypothesis probability x based on condition y

(posteriori probability)

P (x): hypothesis probability x (prior probability)

P (y|x): probability y based on the conditions in

hypothesis

x p (y): probability to y

The largest probability value belongs to the

appropriate class. As a classify data, it only requires

all the possibilities that occur. Naïve Bayes is an

algorithm that is included in supervised learning so

the initial learning process is needed to make

decisions. The classification process with Naïve

Bayes is done using the training data that was

previously divided using K-fold cross validation. In

order to conduct the learning or testing the character

data will be taken one by one from the previously

existing features. At this stage of classification there

are two processes: Learning level using existing data,

the second is estimate the parameters of the

distribution of the opportunity with the assumption

that the independence of each class (data with the

same characteristics). In this stage is estimated in the

parameters with Maximum Likelihood (ML), and the

predictive stage is the process of using the model that

has been built to conduct data tests to

estimate/measure the accuracy of the rules formed in

the model by calculating the opportunity posterior

Design of Alcohol Detection and Classiﬁcation Devices in Traditional Legen / Tuak Drinks using an IoT-based MQ-3 Sensor

279

then classify into the largest opportunity posterior

MAPH (Maximum A Posteriori Hypothesis)

(Manalu, Sianturi, & Manalu, 2017).

3 IMPLEMENTATION

3.1 System Implementation

Figure 2. system implementation

Figure 2 shows the schematic wiring circuit from

sensors and the output to micro with the following

details:

a. Sensor Ph: V +, G (GND), Po (A1)

b. MQ alcohol Sensors-3: VCC (5v), GND,

c. Temperature Sensor DHT11: Black

(GND), Red (5v), yellow (D2)

d. LCD: VCC (5v), GND, SDA, SCL

e. Node MCU ESP 8266: Vin (5v), GND,

D6 (serial communication)

3.2 Hardware Implementation

Hardware implementation, in this study used 3

sensors, MQ-3 sensors for alcohol content detection,

DS18B20 temperature sensors, pH Module E201-

CBNC for pH levels detection. The Micro

ATmega2560, NodeMCU ESP-8266 as the data

sender to the server. At the end of the results testing

process will be displayed on the LCD (Lyquid crystal

Display) and for realtime data can be observed using

IoT devices connected to the device.

Figure 3. hardware

3.3 IoT (Internet of Things)

As described in the previous section, IoT is used as a

data viewer for processing results from detection and

classification on the smartphones. IoT (Internet of

Things) is the latest communication device, where

electronic devices can be integrated with each other

with microcontroller, wave transmitter for

communication, and good protocol stacks (Ikegami,

1997).

In making the IoT system (Internet of Things)

requires device conection (device connection) and

data Sensing (data sensing). IoT can be said as a

package of things that are interconnected over the

internet consisting of sensors, tags, and others. IoT is

used to collect information and data that can then be

processed. IoT can be applied in the field, such as in

health, agriculture, smart building, transportation,

Smart grid, Automation and others (Winasis,

Nugraha, Rosyadi, & Nugroho, 2016)(Rahmawati,

2019) (Rao, Ajit, & Kumar, 2018)(Bhagwat, Hulloli,

Patil, Khan, & Kamble, 2018).

For the IoT platform in this study using ubidots as

a data receiver server and displaying the results of all

sensors (Hidayatullah, Fat, & Andriani, 2018).

Testing with a serial monitor, Testing a web server

and Testing on Android is well connected.

3.4 Research Results

3.4.1 Using Naïve Bayes Method

Testing by method, Navie Bayes method is a simple

testing process, which less formulation but very

precision. At the test this time begins with

temperature detection, then pH testing and alcohol

testing. Once the ADC voltage reads will be

processed for conversion, the next displays the

percentage or content of each readable on the sensor.

This method begins with collecting the training data.

HIMBEP 2020 - International Conference on Health Informatics, Medical, Biological Engineering, and Pharmaceutical

280

Table 1. Sample Training Data

type of

sample

lab test results

Alkohol

(%)

temperature

(°C)

1A 6,5 1 29-30

1B 6,7 3 29-30

1C 6,7 3 29-30

1D 3,5 4 29-30

1E 3,5 5 29-30

2A 6,8 1 29-30

2B 5,6 4 29-30

2C 3,4 6 29-30

2D 3,3 6 29-30

2E 3,0 6 29-30

The data can be done to the next process of converting

the value. By making the average in the data group.

Table 2. Temperature classification on samples

tempe

rature

(°C)

Classification

cold Medium Hot

<20 >20, <30 >30

Temperature classification is based on the number

of temperature percentages in degrees Celsius (°c),

"cold" for temperatures below or less than 20 °c,

"moderate" classifications for temperatures greater

than 20 °c and less than 30 °c and "heat" for

temperatures over 30 ° C. Temperature classification

is required to know.

Table 2. Classification of alcohol on samples

Alcohol

(%)

Classification

good Good

enou

Dangerous

<1 >1,<5 >5

Classification on alcohol content is done based on

the number of percentage of alcohol has been read,

the alcohol content is read in percentages (%) If the

alcohol level in the sample is read below 1% then the

condition can be said "good" in the sense of well

worth the consumption, for a percentage above 1%

and less than 5% then the condition "good enough ".

Table 3. pH Classification on samples

pH Classification

sour Normal Wet

4,00 >5,86 <9,18

Classification on pH content, in general, the drink

is considered worthy to be consumed ie that has a

magnitude of pH > 6.86 and < 9.18 but different for

the sample used this time for the condition "Normal"

content on the pH must be in the range of < 9.18 and

> 6.86 + 1% tolerances so the sample can be said to

Normal until pH 5.86 for the condition "sour" pH < 4

or < 5.86.

Table 4. Conversion of classification values

Information Alcohol

(%)

pH temperature

(°C)

GOOD <1 >5,86,

<9,18

>20,<30

GOOD

ENOUGH

>1, <5 4.00 <20

DANGEROUS >5 <9,18 >30

From the table then the training data can be

narrated or classified as follows:

Table 5. Data Trining Classification

Sample

Type

lab test results

pH Alkohol temperat

ure

Information

1A Normal GOOD Medium GOOD

1B Normal Good Enough Medium GOOD

ENOUGH

1C Normal Good Enough Medium GOOD

ENOUGH

1D Sour Good Enough Medium GOOD

ENOUGH

1E Sour Dangerous Medium DANGEROUS

2A Normal Good Medium GOOD

2B Sour Good Enough Medium GOOD

ENOUGH

2C Sour Dangerous Medium DANGEROUS

2D Sour Dangerous Medium DANGEROUS

2E Sour Dangerous Medium DANGEROUS

For data testing can be written with "Data 11" as

in the following data:

Table 6. Data testing

Sampel Ph Alkohol

(%)

temperature

(°C)

Information

3A Normal Goo

Medium ?

In data 11, this data testing obtained Normal pH

result, alcohol content is still good and with moderate

temperature. Results on the caption will be searched

using Navie Bayes:

Design of Alcohol Detection and Classiﬁcation Devices in Traditional Legen / Tuak Drinks using an IoT-based MQ-3 Sensor

281

Step 1

The appearance of "GOOD" in the description

data is 2 items

The appearance of "GOOD ENOUGH" in the

description data is 4 items

Occurrence "DANGEROUS" in the description

data is 4 items

Then,

C “GOOD” = 2/10

P|C “GOOD ENOUGH” = 4/10

C “DANGEROUS” = 4/10

Step 2

The second step is data collection where data

collection is the calculation of each other's

information.

Calculate the pH of "Normal" which has the

description "GOOD", "GOOD ENOUGH",

"DANGEROUS".

ormal

GOOD =2/2

ormal

GOOD ENOUGH =2/4

ormal|DANGEROUS =0/4

Calculate alcohol "Good" which has the

description "GOOD", "GOOD ENOUGH",

"DANGEROUS".

Good|GOOD =2/2

Good|GOOD ENOUGH =0/4

Good|DANGEROUS =0/4

Calculate the temperature "Medium" which has

the description "GOOD", "GOOD ENOUGH",

"DANGEROUS".

Medium

GOOD =2/2

Medium|GOOD ENOUGH =2/4

Medium

DANGEROUS =4/4

Step 3

The third step is data collection, where data

collection is classified into the calculation of

information from each.

"GOOD" classification

P|C “GOOD” =2/10

ormal|GOOD =2/2

Good|GOOD =2/2

Medium|GOOD =2/2

Then, P | GOOD = 2/10 * 2/2 * 2/2 * 2/2

P | GOOD = 0.2

The "GOOD ENOUGH" classification

C “GOOD ENOUGH” =4/10

ormal

GOOD ENOUGH =2/4

Good|GOOD ENOUGH =0/4

Medium

GOOD ENOUGH =2/4

Then, P | ENOUGH GOOD = 4/10 * 2/4 * 0/4

* 2/4

P | GOOD ENOUGH = 0

"DANGEROUS" classification

C “DANGEROUS” =4/10

ormal|DANGEROUS =0/4

Good

DANGEROUS =0/4

Medium

DANGEROUS =4/4

Then, P | DANGEROUS = 4/10 * 0/4 * 0/4 *

4/4

P | DANGEROUS = 0

3.4.2 Classification Results

In this final result a comparison of GOOD, GOOD

ENOUGH, and DANGER results from the

calculations done in step 1, step 2 and step 3. The final

result is determined from the magnitude of the

comparison value of the three (3) classifications,

which results are greater:

GOOD = 0,2

ormal|GOOD ENOUGH = 0

Good

DANGEROUS = 0

So it can be concluded from the data above, the data

which is rated greater is "GOOD" then the results for

the answer to the test data (data11) in search are

"GOOD".

Table 7. Testing data results

Sample Ph

Alkohol

(%)

temperature

(°C)

Keterang

3A Normal Good Medium GOOD

Table 8. Trial Results for the entire sample

HIMBEP 2020 - International Conference on Health Informatics, Medical, Biological Engineering, and Pharmaceutical

282

From the test results in table 9. it can be seen the

contribution of success / accuracy by the Navie Bayes

method using data classification.

Table 9. Classification of data captions

NO Naïve Bayes

results

Tool Test

Results

information

1 GOOD GOOD Corresponding

2 GOOD

ENOUGH

GOOD

ENOUGH

Corresponding

3 GOOD

ENOUGH

GOOD

ENOUGH

Corresponding

4 GOOD

ENOUGH

GOOD

ENOUGH

Corresponding

5 DANGEROUS GOOD

ENOUGH

Not

Corresponding

6 GOOD GOOD Corresponding

7 GOOD

ENOUGH

GOOD

ENOUGH

Corresponding

8 DANGEROUS DANGEROUS Corresponding

9 DANGEROUS DANGEROUS Corresponding

10 DANGEROUS DANGEROUS Corresponding

Of the 10 data that are owned, which have

information that does not match one (1) data and that

has information according to as many as nine (9) data,

the level of accuracy of tool testing using this method

is very good.

100%  90%

The level of accuracy is 90% from 100%.

4 CONCLUSION

The authenticity of a Legen sample/Tuak (90-100%)

Can be known using this tool. For Legen testing/Tuak

that is not genuine gas alcohol properties in the

Legen/Tuak only large on the gas is not on the drink

is proven when the long left in the air the alcohol

content that reads very minimal is different from the

original it is compared.

The Navie Bayes method used as a sample

classification method is very well proven by

achieving a 90% success rate on tool testing. The

connected IoT system is excellent showing the work

of each sensor in realtime.

REFERENCES

Alkohol, U. J. I., & Fermentasi, P. (n.d.). Uji alkohol pada

fermentasi tuak. 148–156.

Artikel, I. (2018). Jurnal Litbang Industri. 23–30.

Ayu, I. G., Dhyanaputri, S., & Jirna, I. N. (2017). KAJIAN

KARAKTERISTIK OBJEKTIF DAN SUBJEKTIF

TUAK AREN ( Arenga pinnata ) BERDASARKAN

LAMA WAKTU PENYIMPANAN. 5(2).

Ayu, I., Pranayanti, P., & Sutrisno, A. (2015).

PEMBUATAN MINUMAN PROBIOTIK AIR

KELAPA MUDA ( Cocos nucifera L . ) DENGAN

STARTER Lactobacillus casei strain Shirota The

Making of Coconut Water ( Cocos nucifera L .)

Probiotic Drink with Starter Lactobacillus casei Shirota

strain. 3(2), 763–772.

Bhagwat, S. D., Hulloli, A. I., Patil, S. B., Khan, A. A., &

Kamble, A. S. (2018). Smart Green House using IOT

and Cloud Computing. 2330–2333.

Bhuta, P., Desai, K., & Keni, A. (2015). Alcohol Detection

and Vehicle Controlling. 2015 International Journal of

Enginering Trends and Applications (IJETA), 2(2), 92–

97.

Fatmawati, F. (2016). Perbandingan Algoritma Klasifikasi

Data Mining Model C4.5 Dan Naive Bayes Untuk

Prediksi Penyakit Diabetes. None, 13(1), 50–59.

Hidayatullah, M., Fat, J., & Andriani, T. (2018). Prototype

Sistem Telemetri Pemantauan Kualitas Air Pada Kolam

Ikan Air Tawar Berbasis Mikrokontroler. Positron,

8(2), 43. https://doi.org/10.26418/positron.v8i2.27367

Ikegami, S. (1997). Tuak dalam Masyarakat Batak Toba:

Laporan Singkat tentang Aspek Sosial-budaya

Penggunaan Nira Tuak in the Toba Batak Society: A

Preliminary Report on the Socio-cultural Aspect of

Palm Wine Consumption. (11), 1–8.

Ilmu, D., & Dalam, P. (2019). Fermentasi: Metode untuk

Meningkatkan Nilai Nutrisi Jerami Padi Fermentation:

Methods to Improve Nutrition Value of Rice Straw

Yanuartono, S. Indarjulianto, H. Purnamaningsih, A.

Nururrozi, dan S. Raharjo. 14(1), 49–60.

Lubis, R. S., & Pinem, M. (2014). Analisis Quality of

Service (Qos) Jaringan Internet Di Smk Telkom

Medan. Singuda ENSIKOM, 7(3), 131–136.

Manalu, E., Sianturi, F. A., & Manalu, M. R. (2017).

Penerapan Algoritma Naive Bayes Untuk Memprediksi

Jumlah Produksi Barang Berdasarkan Data Persediaan

Dan Jumlah Pemesanan Pada Cv . Papadan Mama

Pastries. Effrida Manalu, Fricles Ariwisanto Sianturi,

Mamed Rofendy Manalu, 1(2), 16–21.

Moede, F. H., & Gonggo, S. T. (2017). PENGARUH

LAMA WAKTU FERMENTASI TERHADAP

KADAR BIOETANOL DARI PATI UBI JALAR

KUNING ( Ipomea batata L ) The Influence of A Long

Time Fermentation Againts bioethanol levels of Starch

Sweet Potato is Yellow ( Ipomea batatas L ). 6(May),

86–91.

Pamungkas, W., & Kompiang, M. (2006). Teknologi

fermentasi, alternatif solusi dalam upaya pemanfaatan

bahan pakan lokal. 43–48.

Design of Alcohol Detection and Classiﬁcation Devices in Traditional Legen / Tuak Drinks using an IoT-based MQ-3 Sensor

283

Pangan, M. T., & Pelita, U. (2017). , R Montana 1 , NC

Putri 1. 3(2), 105–116.

Profile, S. E. E. (2018). Classification Using Naïve Bayes

and Decision Tree on Food Addiction Classification

Using Naïve Bayes and Decision Tree on Food

Addiction. (January 2016).

https://doi.org/10.14257/ijdta.2016.9.3.17

Rahmawati, D. (2019). Pengujian Monitoring On-Line

Rumah Kaca Cerdas Berbasis Android. Cyclotron, 2(1).

https://doi.org/10.30651/cl.v2i1.2529

Rao, M. N., Ajit, K. C., & Kumar, G. P. (2018). Smart

Green House Based on IOT. 7, 258–261.

Raschka, S. (2014). Introduction and Theory. 1–20.

Rizal, S., Erna, M., Nurainy, F., & Tambunan, R. (2016).

Karakteristik Probiotik Minuman Fermentasi Laktat

Sari Buah Nanas dengan Variasi Jenis Bakteri Asam

Laktat Probiotic Characteristic of Lactic Fermentation

Beverage of Pineapple Juice with Variation of Lactic

Acid Bacteria ( LAB ) Types mengonsumsi minuman.

18(June), 63–71.

Winasis, W., Nugraha, A. W. W., Rosyadi, I., & Nugroho,

F. S. T. (2016). Desain Sistem Monitoring Sistem

Photovoltaic Berbasis Internet of Things (IoT). Jurnal

Nasional Teknik Elektro Dan Teknologi Informasi

(JNTETI), 5(4), 328–333.

https://doi.org/10.22146/jnteti.v5i4.281

HIMBEP 2020 - International Conference on Health Informatics, Medical, Biological Engineering, and Pharmaceutical

284