Fungsionalisasi Electronic Nose untuk Mendiferensiasi Rambak Kulit dari Babi, Kambing, dan Kerbau yang Dikombinasikan dengan Machine Learning

Hambali Putra Perkasa, Prof. Dr. Eng. Kuwat Triyana, M.Si.

2024 | Skripsi | FISIKA

Penelitian ini mengaplikasikan electronic nose (e-nose) yang dilengkapi delapan sensor gas berbasis semikonduktor untuk mendeteksi dan membedakan jenis rambak berdasarkan bahan bakunya, yaitu kulit babi, kambing, dan kerbau. Identifikasi jenis rambak penting dilakukan karena secara visual bahan baku rambak sulit dibedakan. Metode autentikasi konvensional seperti Gas Chromatography-Mass Spectrometry (GC-MS) dan Fourier Transform Infrared (FTIR) memiliki keterbatasan berupa waktu analisis yang lama, biaya mahal, dan membutuhkan operator ahli. Sampel rambak kulit mentah tanpa bumbu diperoleh dari tiga produsen dengan total 50 data pengukuran berulang untuk setiap jenis bahan baku. Pengujian dilakukan pada suhu 60°C tanpa variasi suhu. Data respon sensor diproses dan dianalisis menggunakan Principal Component Analysis (PCA) untuk klasterisasi awal dan machine learning berbasis Linear Discriminant Analysis (LDA) untuk klasifikasi. Data dipecah menjadi 80% untuk pelatihan dan 20% untuk pengujian dengan metode k-fold cross-validation guna memastikan konsistensi model. Hasil penelitian menunjukkan bahwa prototipe e-nose berhasil mengklasifikasikan jenis rambak berdasarkan bahan bakunya, dengan tingkat variabilitas klasterisasi sebesar 72,41% menggunakan PCA dan akurasi pengujian model sebesar 90% menggunakan LDA. Penelitian ini membuktikan bahwa e-nose dapat menjadi alat autentikasi yang efisien, hemat biaya, non-destruktif, dan mudah digunakan, khususnya untuk produk berbasis bahan baku kulit. 

This study aims to apply an electronic nose (e-nose) equipped with eight semiconductor-based gas sensors to detect and differentiate types of rambak (crackers) made from pig, goat, and buffalo skin. Identifying the raw materials of rambak is crucial, as they are visually indistinguishable. Conventional authentication methods such as Gas Chromatography-Mass Spectrometry (GC-MS) and Fourier Transform Infrared (FTIR) spectroscopy are effective but have limitations, including long processing times, high costs, and the requirement for skilled operators. Raw rambak samples were commercially obtained from three different producers, with each type of rambak providing 50 repeated measurements using the e-nose. The samples were unseasoned and tested at a constant temperature of 60°C. Sensor response data were processed and analyzed using Principal Component Analysis (PCA) for clustering and Linear Discriminant Analysis (LDA) for classification. The dataset was divided into 80% for training and 20% for testing, with k-fold cross-validation employed to ensure model consistency. The results showed that the e-nose prototype successfully classified the rambak types based on their raw materials, achieving 72.41% variability in clustering using PCA and a 90% testing accuracy using LDA. These findings demonstrate that the e-nose, combined with advanced pattern recognition techniques, can serve as an efficient, cost-effective, non-destructive, and user-friendly detection tool.

Kata Kunci : Electronic nose, rambak kulit, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), autentikasi bahan baku