KLASIFIKASI KUALITAS TELUR AYAM BERDASARKAN FITUR AUDIO MENGGUNAKAN DEEP LEARNING

DELILA SEPTIANI DWI PUTRI, . (2024) KLASIFIKASI KUALITAS TELUR AYAM BERDASARKAN FITUR AUDIO MENGGUNAKAN DEEP LEARNING. Sarjana thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (1MB)
	Text Bab 1.pdf Download (463kB)
	Text Bab 2.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text Bab 3.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text Bab 4.pdf Restricted to Registered users only Download (992kB) | Request a copy
	Text Bab 5.pdf Restricted to Registered users only Download (485kB) | Request a copy
	Text Daftar Pustaka.pdf Download (591kB)
	Text Lampiran.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Telur ayam merupakan salah satu produk yang mudah rusak karena berbagai faktor, sehingga tak jarang konsumen mendapati telur dengan kualitas tidak layak konsumsi. Pada penelitian ini, kualitas telur ayam diklasifikasikan menggunakan model deep learning berdasarkan fitur audio dalam domain waktu dan frekuensi yang diekstrak dari suara ketukan telur. Model yang digunakan meliputi Deep Neural Network (DNN) dan Convolutional Neural Network (CNN). Eksperimen dilakukan dengan menggunakan sampel 200 telur ayam negeri yang terdiri atas telur segar (layak konsumsi) dan busuk (tidak layak dikonsumsi). Data suara diperoleh dengan mengetuk telur secara vertikal dan horizontal di dalam sebuah kotak kayu tertutup berukutan 39.5 ×28×21.5 cm. Adapun fitur yang diekstraksi meliputi energy, entropy, Zero Crossing Rate (ZCR), spectral centroid, dan MFCC. Data fitur audio dianalisis dengan uji statistik Mann-Whitney U serta Principal Component Analysis (PCA) dan clustering. Fitur audio yang memiliki p-value <0.05 akan dijadikan input dalam model klasifikasi. Hasil penelitian menunjukkan bahwa model CNN dengan input fitur MFCC memiliki akurasi terbaik dalam penelitian ini, yaitu sebesar 98.75%. Di samping itu, model DNN dengan input fitur dalam domain waktu dan frekuensi yang diperoleh pada pengetukan horizontal dan vertikal masing-masing memiliki akurasi sebesar 90.00% dan 92.50%.Dengan demikian, penelitian ini dapat dikembangkan lebih lanjut sebagai solusi dalam mengatasi penjaminan kualitas bahan pangan yang akan dikonsumsi masyarakat.*****Chicken eggs are one of the products that spoil easily due to various factors, so it is not uncommon for consumers to find eggs that are not fit for consumption. In this research, the quality of chicken eggs is classified using deep learning based on audio features in the time and frequency domains extracted from the sound of tapping the eggs. Models used include Deep Neural Network (DNN) and Convolutional Neural Network (CNN). Experiments were conducted using a sample of 200 chicken eggs consisting of fresh and rotten eggs (suitable and unsuitable for consumption). The audio data was obtained by tapping the eggs vertically and horizontally inside a closed wooden box measuring 39.5 ×28×21.5 cm. The extracted features included energy, entropy, Zero Crossing Rate (ZCR), spectral centroid, and MFCC. The audio feature data was analyzed using the Mann-Whitney U statistical test as well as Principal Component Analysis (PCA) and clustering. Audio features with a p-value < 0.05 were used as input in the classification models. The research results showed that the CNN model with MFCC feature input had the best accuracy in this study, which was 98.75%. Additionally, the DNN model with input features in the time and frequency domains obtained from horizontal and vertical tapping achieved accuracies of 90.00% and 92.50%, respectively. Thus, this research can be further developed as a solution to ensure the quality assurance of food products to be consumed by the public.

Item Type:	Thesis (Sarjana)
Additional Information:	1). Dr.rer.nat Bambang Heru Iswanto, M.Si. ; 2). Haris Suhendar, S.Si., M.Sc.
Subjects:	Sains > Fisika
Divisions:	FMIPA > S1 Fisika
Depositing User:	Users 24903 not found.
Date Deposited:	20 Aug 2024 01:24
Last Modified:	20 Aug 2024 01:24
URI:	http://repository.unj.ac.id/id/eprint/50141

Actions (login required)

View Item