Wijaya, Dedy Rahman (2019) Mobile Electronic Nose Framework untuk Deteksi Kualitas Daging Sapi. Doctoral thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of Upload_Perpus.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
Upload_Perpus.pdf Restricted to Repository staff only Download (24MB) | Request a copy |
Abstract
Saat ini, kasus keracunan makanan masih banyak terjadi di berbagai negara di dunia. Di Amerika Serikat, sekitar 76 juta kasus keracunan makanan terjadi setiap tahun. Kerugian ekonomi yang disebabkan oleh penyakit dari patogen utama diperkirakan lebih dari USD 35 miliar untuk biaya medis dan kehilangan produktivitas setiap tahunnya. Otoritas keamanan pangan Eropa juga telah melaporkan bahwa lebih dari 15% populasi di Eropa menderita penyakit bawaan makanan. Sedangkan di Indonesia, terdapat 1068 kasus keracunan makanan pada tahun 2016. Sementara itu, daging sapi merupakan salah satu sumber animal-based protein (ABP) yang paling populer untuk dikonsumsi. Kementrian Pertanian Republik Indonesia memproyeksikan rata-rata permintaan daging sapi tahun 2013-2019 tumbuh sebesar 0.86% per tahun. Sedangkan Badan Pangan Dunia (Food and Agriculture Organization) merilis laporan dan proyeksi konsumsi daging di dunia akan terus mengalami kenaikan hingga tahun 2050. Pada kenyataannya, daging sapi adalah bahan makanan yang mudah rusak jika tidak ditangani dengan baik. Daging merupakan media yang ideal untuk pertumbuhan mikroba. Mikroba khususnya bakteri berkembang biak dengan mengurai nutrisi yang ada pada daging. Penyimpanan daging yang kurang baik seperti penyimpanan pada suhu kamar dan di udara terbuka akan mempercepat degradasi kualitas daging sapi karena merupakan lingkungan paling optimal untuk pertumbuhan bakteri. Kondisi seperti inilah yang sering ditemukan pada penjual daging terutama di Indonesia.
Dengan melihat penjelasan di atas maka diperlukan adanya sistem deteksi kualitas daging sapi yang murah dan mudah digunakan. Pada penelitian beberapa tahun terakhir, electronic nose (e-nose) digunakan untuk deteksi kualitas daging sapi untuk membedakan daging menjadi dua kelas (segar dan busuk) serta tiga kelas (segar, agak busuk, dan busuk). Pada penelitian ini, e-nose digunakan untuk deteksi daging sapi multikelas dengan yang terdiri empat kelas (excellent, good, acceptable, spoiled). Selain itu, prediksi populasi mikroba juga dilakukan menggunakan metode regresi. Penelitian ini terdiri dari lima tahap utama sebagai berikut: signal preprocessing, sensor array optimization, classification/regression technique, sensing as a service (S2aaS) framework, dan pengembangan mobile application. Kontribusi dari penelitian ini antara lain adalah usulan metode noise filtering, metode optimasi sensor array, metode klasifikasi daging sapi multiclass termasuk prediksi populasi mikroba pada sampel daging, serta prototype dan framework S2aaS untuk proses analisis data e-nose secara online. Sedangkan kontribusi praktis dari penelitian adalah mengembangkan sistem uji/deteksi kualitas daging menggunakan mobile electronic nose yang akurat, cepat, dan mudah digunakan. Hasil eksperimen menunjukkan bahwa e-nose mampu mengklasifikasikan empat kelas daging dengan rata-rata akurasi lebih dari 95%. Sedangkan, hasil prediksi mikroba memiliki rata-rata MSE hanya 0.0631. Selain itu, MoLenNet sebagai framework S2aaS telah diusulkan dan berhasil diimplementasikan.
================================================================================================
At present, food poisoning cases still occur in many countries in the world. In the United States, around 76 million cases of food poisoning occur every year. Economic losses caused by this illness are more than USD 35 billion for medical expenses and productivity losses each year. European Food Safety Authority has been reported that more than 15% of the populations in Europe suffering from foodborne diseases. In Indonesia, there were 1068 cases of food poisoning in 2016. On the other side, beef is one of the most popular sources of animal-based protein (ABP). The Indonesian Ministry of Agriculture estimates that the average demand for beef in 2013-2019 will grow by 0.86% per year. World Food and Agriculture Organization also released reports of meat consumption in the world will continue to increase until 2050. In fact, beef is kind of perishable food if not handled properly. Actually, meat is an ideal medium for microbial growth. Microbes, especially bacteria, multiply by breaking down the nutrients in meat. Poor storage of meat such as storage at room temperature and open air will accelerate the degradation of beef quality because it is the most optimal environment for bacterial growth. This condition is often found in many meat sellers, especially in Indonesia.
Based on the above explanation, it is necessary to develop a beef quality detection system that is rapid and easy to use. In recent years, electronic nose (e-nose) is used to detect the quality of beef to distinguish meat into two classes (fresh and spoiled) and three classes (fresh, semi-fresh, and spoiled). In this study, e-nose was used for detection of multiclass beef quality with four classes (excellent, good, acceptable, spoiled). In addition, prediction of microbial populations was also carried out by regression methods. This study consisted of the following five main stages: signal preprocessing, sensor array optimization, classification/regression technique, sensing as a service (S2aaS) framework, and mobile application development. Contributions of this study including noise filtering method, sensor array optimization method, multiclass beef classification method as well as microbial population prediction on beef samples, S2aaS prototype and framework for online e-nose data analysis. Moreover, the practical contribution of the research is to develop a meat quality detection system using a mobile electronic nose that is accurate, low-cost, and easy to use for personal or household scales. The experimental results show that the e-nose is able to classify four classes of meat with an average accuracy of more than 95%. Moreover, microbial prediction results have an average MSE of only 0.0631. In addition, MoLenNet as a S2aaS framework has been proposed and successfully implemented.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Additional Information: | RDIf 004 Wij m-1 2019 |
| Uncontrolled Keywords: | mobile electronic nose, deteksi kualitas daging sapi, multiclass classification, regression |
| Subjects: | Q Science > QA Mathematics > QA76.585 Cloud computing. Mobile computing. T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55001-(S3) PhD Thesis |
| Depositing User: | Dedy Rahman Wijaya |
| Date Deposited: | 23 May 2022 02:24 |
| Last Modified: | 23 May 2022 02:24 |
| URI: | http://repository.its.ac.id/id/eprint/62761 |
Actions (login required)
 |
View Item |