Electronic Nose (E-Nose) sebagai Detektor Formalin pada Bahan Pangan Menggunakan ANN Dilengkapi Feature Extraction Berbasis Principal Component Analysis (PCA)

Anandhita, Firda (2026) Electronic Nose (E-Nose) sebagai Detektor Formalin pada Bahan Pangan Menggunakan ANN Dilengkapi Feature Extraction Berbasis Principal Component Analysis (PCA). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2042221108-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Penyalahgunaan formalin pada bahan pangan, khususnya tahu, masih menjadi permasalahan karena keberadaannya sulit dikenali secara visual. Penelitian ini bertujuan merancang sistem electronic nose (e-nose) sebagai detektor formalin pada tahu serta mengevaluasi performanya menggunakan Artificial Neural Network (ANN) dengan feature extraction berbasis Principal Component Analysis (PCA). Sistem e-nose menggunakan Gas Sensor Array (GSA) yang terdiri atas sensor Grove HCHO, MQ-138, dan TGS822, serta sensor DHT22 untuk mengukur kelembapan dan DS18B20 untuk memantau temperatur air pada bak pemanas. Pengujian dilakukan menggunakan sampel tahu seberat 50 gram dengan variasi penambahan larutan formalin 37% sebesar 0, 1, 5, 10, dan 15 mL, dengan masing-masing variasi dilakukan sebanyak lima replikasi sehingga diperoleh 25 sampel. Data respons sensor diekstraksi menjadi 13 fitur statistik, kemudian distandardisasi dan ditransformasikan menggunakan PCA. Tiga komponen utama, yaitu PC1, PC2, dan PC3, digunakan sebagai fitur masukan pada model ANN. Evaluasi model dilakukan menggunakan Leave One-Replication-Out Cross-Validation (LORO-CV). Hasil menunjukkan bahwa tiga komponen utama mampu mempertahankan 91,27% variasi data. Model PCA-ANN berhasil mengklasifikasikan 24 dari 25 sampel dengan benar dan memperoleh accuracy sebesar 96%, precision 100%, recall 95%, specificity 100%, F1-score 97,14%, serta ROC-AUC 100%. Model yang telah dilatih kemudian diimplementasikan pada Raspberry Pi untuk melakukan klasifikasi secara real-time. Hasil penelitian menunjukkan bahwa sistem e-nose berbasis PCA-ANN mampu membedakan sampel tahu formalin dan non-formalin dengan performa klasifikasi yang baik.
==================================================================================================================================
The misuse of formaldehyde in food products, particularly tofu, remains a concern because its presence is difficult to identify visually. This study aims to design an electronic nose (e-nose) system as a formaldehyde detector for tofu and to evaluate its performance using an Artificial Neural Network (ANN) with feature extraction based on Principal Component Analysis (PCA). The e-nose system employs a Gas Sensor Array (GSA) consisting of Grove HCHO, MQ-138, and TGS822 sensors, along with a DHT22 sensor for humidity measurement and a DS18B20 sensor for temperature monitoring. The experiments were conducted using 50 gram tofu samples with formalin 37% addition levels of 0, 1, 5, 10, and 15 mL, with each variation tested in five replications, resulting in a total of 25 samples. The sensor response data were extracted into 13 statistical features, followed by standardization and transformation using PCA. Three principal components, namely PC1, PC2, and PC3, were used as input features for the ANN model. Model evaluation was performed using Leave-One-Replication-Out Cross-Validation (LORO-CV). The results showed that the three principal components retained 91.27% of the total data variance. The PCA-ANN model correctly classified 24 out of 25 samples, achieving an accuracy of 96%, precision of 100%, recall of 95%, specificity of 100%, F1-score of 97.14%, and ROC-AUC of 100%. The trained model was subsequently implemented on a Raspberry Pi for real-time classification. The results demonstrate that the PCA-ANN-based e-nose system is capable of distinguishing formaldehyde-treated and non formaldehyde-treated tofu samples with good classification performance.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Artificial Neural Network, electronic nose, formalin, gas sensor array, Principal Component Analysis. Artificial Neural Network, electronic nose, formaldehyde, gas sensor array, Principal Component Analysis.
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing T Technology > TA Engineering (General). Civil engineering (General) > TA1573 Detectors. Sensors
Divisions:	Faculty of Vocational > Instrumentation Engineering
Depositing User:	FIRDA ANANDHITA
Date Deposited:	14 Sep 2026 12:40
Last Modified:	25 Sep 2026 03:23
URI:	http://repository.its.ac.id/id/eprint/144478

Actions (login required)

View Item