Pengaruh Jumlah Sensor Electronic Nose Untuk Mengklasifikasi Kandungan Formaldehida Pada Cumi-Cumi

Kurniadewi, Kadek Nesya (2022) Pengaruh Jumlah Sensor Electronic Nose Untuk Mengklasifikasi Kandungan Formaldehida Pada Cumi-Cumi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05311840000009-Undergraduate_Thesis.pdf Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Pengembangan alat electronic nose dapat memudahkan untuk melakukan deteksi terhadap berbagai jenis gas. Untuk memenuhi fungsinya, electronic nose dikembangkan dengan susunan dari beberapa sensor dan didukung dengan microcontroller Arduino. Selain melakukan deteksi, pada penelitian kali ini penulis akan melakukan perbandingan hasil akurasi sensor MQ pada saat melakukan klasifikasi kandungan formalin pada cumi-cumi sesuai dengan kelasnya yaitu kelas [0,1,2,3,4,5], masing masing kelas berarti kandungan formalinnya sebanyak [0%, 1%, 5%, 10%, 20%, 30%]. Penelitian ini berfokus untuk mengetahui pengaruh jumlah dan jenis sensor MQ yang digunakan terhadap hasil akurasi klasifikasi menggunakan algoritma Neural Network (NN). Pada electronic nose terdapat sensor array yang terdiri dari 6 jenis sensor MQ (MQ2, MQ3, MQ7, MQ8, MQ135, MQ138) yang akan digunakan untuk mendeteksi gas terhadap kadar formalin pada cumi-cumi. Cumi-cumi digunakan sebagai objek penelitian dengan 6 jenis kandungan formaldehida yang berbeda saat pengambilan data dilakukan dengan memanfaatkan alat electronic nose. Data yang didapatkan dari hasil deteksi menggunakan electronic nose kemudian dipre-proses untuk diambil bagian sedot objek saja (P2), kemudian dihitung rata-rata setiap 1 file pengambilan data, kemudian data rata-rata tersebut di-reduksi menggunakan Principal Component Analysis (PCA). Algoritma Neural Network (NN) dilatih untuk dapat menentukan ground truth dari nilai setiap sensor pada suatu label untuk mengklasifikasikan kandungan formalin berdasarkan kelasnya masing-masing dengan hasil akurasi sebesar 96%. Menentukan jumlah dan jenis sensor terbaik dilakukan dengan mencari kelompok sensor yang memiliki akurasi tertinggi saat diklasifikasi menggunakan algoritma Neural Network (NN), dengan hasil sensor terbaik yaitu MQ2, MQ7, MQ135, dan MQ138 dengan akurasi sebesar 99%. Dilakukannya uji analisis sensor tersebut pada algoritma Support Vector Machine (SVM), K-Nearest Neighbor (KNN), dan Decision Tree (DT) dengan akurasi masing-masing 99%, 100%, 99%.
====================================================================================================================================
The development of electronic nose tools can make it easier to detect various types of gases. To fulfill its function, the electronic nose was developed with an arrangement of several sensors and supported by an Arduino microcontroller. In addition to detecting, in this study, the author will compare the results of the accuracy of the MQ sensor when classifying the formalin content in squid according to its class, namely [0,1,2,3,4,5], each class means that the formalin content is [0%, 1%, 5%, 10%, 20%, 30%]. This study focuses on determining the influence of the number and type of MQ sensors used on the results of classification accuracy using the Neural Network (NN) algorithm. In the electronic nose, there is an array sensor consisting of 6 types of MQ sensors (MQ2, MQ3, MQ7, MQ8, MQ135, and MQ138) which will be used to detect gas against formalin levels in squid. Squid is used as a research object with 6 different types of formaldehyde content when data collection is carried out by utilizing an electronic nose tool. The data obtained from the detection results using an electronic nose is then pre-processed to take part of the suction of the object only (P2), then calculated on average every 1 data retrieval file, is reduced using Principal Component Analysis (PCA). Neural Network (NN) algorithms are trained to be able to determine the ground truth of the value of each sensor on a label to classify formalin content based on their respective classes with an accuracy result of 96%. Determining the number and type of the best sensors is done by looking for the sensor group that has the highest accuracy when classified using the Neural Network (NN) algorithm, with the best sensor results, there are MQ2, MQ7, MQ135, and MQ138 with an accuracy of 99%. The sensor analysis test was carried out on the Support Vector Machine (SVM), K-Nearest Neighbor (KNN), and Decision Tree (DT) algorithms with an accuracy of 99%, 100%, and 99%, respectively.

Item Type:	Thesis (Other)
Additional Information:	RSTI 621.381 53 Kur p-1 2022
Uncontrolled Keywords:	Formalin, Electronic Nose, Sensor MQ, Neural Network (NN), Deteksi, Akurasi, Klasifikasi, Formaldehyde, Electronic Nose, MQ Sensors, Neural Network (NN), Detection, Accuracy, Classification.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7878 Electronic instruments
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Mr. Marsudiyana -
Date Deposited:	22 Apr 2026 05:44
Last Modified:	22 Apr 2026 05:44
URI:	http://repository.its.ac.id/id/eprint/132859

Actions (login required)

View Item