Autentikasi Penciri Khas Single Origin Biji Kopi Indonesia Berbasis Karakteristik Spektral Laser-Induced Breakdown Spectroscopy (LIBS) dan Analisis Kemometrik

Syahid, Ahmad Miqdam (2025) Autentikasi Penciri Khas Single Origin Biji Kopi Indonesia Berbasis Karakteristik Spektral Laser-Induced Breakdown Spectroscopy (LIBS) dan Analisis Kemometrik. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009211107-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Proses sertifikasi kualitas biji kopi di Indonesia memerlukan metode yang mudah dan biaya yang murah. Kualitas biji kopi dapat terlihat dari single origin-nya. Metode spektroskopi dan analisis kemometrik dapat menyelesaikan masalah tersebut. Akan tetapi, masih ada beberapa kekurangan aplikatif dari metode spektroskopi lain yang telah dilakukan oleh penelitian sebelumnya. Penelitian ini bertujuan untuk mencari penciri khas karakteristik spektral dan metode analisis kemometrik dengan performansi terbaik dalam membedakan single origin biji kopi sangrai Indonesia menggunakan teknik spektroskopi berbasis Laser-Induced Breakdown Spectroscopy (LIBS) dengan bantuan metode augmentasi penambahan gaussian noise dari data interpolasi. Sebelum dilakukan pengukuran instrumen LIBS telah dilakukan kalibrasi kestabilan intensitas spektral terhadap energi laser, kalibrasi pergeseran panjang gelombang oleh efek Doppler, serta optimisasi parameter energi laser dan waktu delay detektor. Data spektral diperoleh dari pengukuran pada sampel single origin asal biji kopi sangrai Indonesia, kemudian dilakukan preprocessing meliputi penyetaraan baseline, smoothing Savitzky-Golay, fitting curve fungsi Lorentz, dan normalisasi rentang [0,1]. Hasil dari analisis PCA didapatkan fingerprint autentikasi single origin biji kopi sangrai Indonesia: variasi intensitas spektral unsur Na yang didukung oleh spektral unsur K, H, dan Ca. Untuk memperkuat akurasi dan keandalan autentikasi dilakukan analisis kemometrik berbasis machine learning: Support Vector Machine (SVM), Partial Least Square Discriminant Analysis (PLS-DA), dan 3D Model k-Nearest Neighbors (3D k-NN). Evaluasi machine learning dilaksanakan dengan rasio training:validating:testing set sebesar 70:20:10. Hasil menunjukkan bahwa metode PLS-DA memberikan akurasi terbaik dengan rata-rata 96,75%, diikuti oleh SVM (96,10%) dan 3D k-NN (87,14%). Penelitian ini menunjukkan bahwa kombinasi LIBS dan analisis kemometrik yang tepat dapat menunjukan akurasi autentikasi single origin asal biji kopi sangrai Indonesia yang optimal. Ditambah lagi, penelitian ini menunjukkan sistem bersifat cepat dengan persiapan sampel yang sederhana sehingga menjadi keunggulan daripada metode spektroskopi yang lainnya.
=====================================================================================================================================
The process of certifying coffee bean quality in Indonesia requires a method that is easy and inexpensive. Coffee bean quality can be seen from its single origin. Spectroscopy and chemometric analysis methods can solve this problem. However, there are still some practical shortcomings of other spectroscopy methods that have been carried out in previous studies. This study aims to identify the distinctive spectral characteristics and chemometric analysis methods with the best performance in distinguishing the single origin of roasted Indonesian coffee beans using Laser-Induced Breakdown Spectroscopy (LIBS) spectroscopy techniques, assisted by the addition of Gaussian noise from interpolated data. Before the LIBS instrument measurements were taken, calibration of spectral intensity stability against laser energy, calibration of wavelength shift due to the Doppler effect, and optimization of laser energy parameters and detector delay time were performed. Spectral data were obtained from measurements on single-origin samples of Indonesian roasted coffee beans, followed by preprocessing including baseline equalization, Savitzky-Golay smoothing, Lorentz curve fitting, and normalization to the range [0,1]. The results of PCA analysis yielded the authentication fingerprint of single-origin roasted Indonesian coffee beans: variations in the spectral intensity of Na supported by the spectra of K, H, and Ca. To enhance the accuracy and reliability of authentication, chemometric analysis based on machine learning was performed: Support Vector Machine (SVM), Partial Least Square Discriminant Analysis (PLS-DA), and 3D k-Nearest Neighbors (3D k-NN) model. Machine learning evaluation was performed with a training:validating:testing set ratio of 70:20:10. The results showed that the PLS-DA method provided the best accuracy with an average of 96.75%, followed by SVM (96.10%) and 3D k-NN (87.14%). This study demonstrates that the combination of LIBS and appropriate chemometric analysis can achieve optimal accuracy in authenticating the single origin of Indonesian roasted coffee beans. Additionally, this study shows that the system is fast with simple sample preparation, making it an advantage over other spectroscopic methods.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Chemometrics Analysis, Data Augmentation, Laser Induced Breakdown Spectroscopy (LIBS), Roasted Coffee Beans, Single Origin, Analisis Kemometrik, Augmentasi Data, Biji Kopi Sangrai Indonesia, Laser Induced Breakdown Spectroscopy (LIBS), Single Origin
Subjects:	Q Science > QC Physics > QC451 Spectroscopy
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Ahmad Miqdam Syahid
Date Deposited:	04 Aug 2025 11:55
Last Modified:	04 Aug 2025 11:55
URI:	http://repository.its.ac.id/id/eprint/125152

Actions (login required)

View Item