KLASIFIKASI KEASLIAN DAN CAMPURAN BAHAN BAKAR MINYAK MENGGUNAKAN SPEKTROFOTOMETER DENGAN METODE MACHINE LEARNING

GREICE SIMBOLON, . (2025) KLASIFIKASI KEASLIAN DAN CAMPURAN BAHAN BAKAR MINYAK MENGGUNAKAN SPEKTROFOTOMETER DENGAN METODE MACHINE LEARNING. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (1MB)
	Text BAB I.pdf Download (106kB)
	Text BAB II.pdf Restricted to Registered users only Download (489kB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (276kB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB V.pdf Restricted to Registered users only Download (72kB) | Request a copy
	Text Daftar Pustaka.pdf Download (228kB)
	Text Lampiran.pdf Restricted to Registered users only Download (503kB) | Request a copy

Abstract

Klasifikasi bahan bakar minyak (BBM) merupakan langkah penting dalam memastikan kualitas dan keaslian BBM di tengah maraknya praktik pencampuran ilegal. Penelitian ini mengusulkan metode klasifikasi BBM berbasis spektrofotometri UV-Vis portabel yang dikombinasikan dengan algoritma machine learning untuk membedakan antara murni RON 92 dan campuran RON 92 RON 90. Sebanyak 180 sampel diuji, terdiri dari Pertamax RON 92 murni dan campuran dengan variasi konsentrasi Pertalite RON 90 (10%-90%). Pengukuran spektrum dilakukan pada rentang 400-900 nm menggunakan sensor CMOS, kemudian data intensitas dikonversi menjadi absorbansi melalui Hukum Beer-Lambert dan di praproses serta normalisasi. Ekstraksi fitur dilakukan dengan dua pendekatan utama, yaitu feature importance berbasis XGBoost pada data absorbansi, serta pada fitur statistik (maksimum, minimum, area di bawah kurva/AUC, dan standar deviasi). Selanjutnya, data dievaluasi menggunakan empat model klasifikasi Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), dan K-Nearest Neighbors (KNN) dengan validasi silang stratified 5-fold dan optimasi hyperparameter melalui GridSearchCV. Hasil penelitian menunjukkan bahwa seluruh model mampu mengidentifikasi BBM murni dan campuran dengan akurasi, presisi, Recall, dan F1-score mendekati 100%, terutama ketika menggunakan fitur absorbansi terpilih. Analisis spektrum juga memperlihatkan adanya pergeseran karakteristik puncak absorbansi pada campuran Pertamax RON 92-Pertalite RON 90 ke panjang gelombang yang lebih tinggi. Dengan demikian, penelitian ini membuktikan bahwa integrasi spektrofotometer portabel dan machine learning sangat efektif untuk klasifikasi dan deteksi keaslian BBM secara cepat, akurat, serta aplikatif di lapangan, sekaligus memberikan kontribusi nyata bagi pengawasan mutu BBM di Indonesia. ***** Classification of fuel oil (BBM) is an important step in ensuring the quality and authenticity of fuel amidst the rampant practice of illegal blending. This study proposes a portable UV-Vis spectrophotometry-based fuel classification method combined with a machine learning algorithm to distinguish between pure RON 92 and blend RON 92 RON 90. A total of 180 samples were tested, consisting of pure Pertamax RON 92 and blends with varying Pertalite RON 90 concentrations (10%-90%). Spectral measurements were carried out in the range of 400-900 nm using a CMOS sensor, then the intensity data was converted to absorbance through the Beer-Lambert Law and preprocessed and normalized. Feature extraction was carried out with two main approaches, namely XGBoost-based feature importance on absorbance data, and on statistical features (maximum, minimum, area under the curve/AUC, and standard deviation). Furthermore, the data was evaluated using four classification models: Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), and K-Nearest Neighbors (KNN) with 5-fold stratified cross-validation and hyperparameter optimization through GridSearchCV. The results showed that all models were able to identify pure and blended fuels with accuracy, precision, recall, and F1-score close to 100%, especially when using selected absorbance features. Spectral analysis also showed a shift in the characteristics of the absorbance peak in the Pertamax RON 92-Pertalite RON 90 mixture to higher wavelengths. Thus, this study proves that the integration of a portable spectrophotometer and machine learning is very effective for the classification and detection of fuel authenticity quickly, accurately, and applicable in the field, while providing a real contribution to fuel quality control in Indonesia.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Dr. rer. nat. Bambang Heru Iswanto, M.Si. ; 2). Fachriza Fathan, M.Si.
Subjects:	Sains > Fisika
Divisions:	FMIPA > S1 Fisika
Depositing User:	Greice Simbolon .
Date Deposited:	20 Aug 2025 08:43
Last Modified:	20 Aug 2025 08:43
URI:	http://repository.unj.ac.id/id/eprint/61675

Actions (login required)

View Item