Please use this identifier to cite or link to this item: https://repositori.uma.ac.id/handle/123456789/26177
Title: Rancang Bangun Alat Pendeteksi Kualitas Minyak Transformator dengan Metode Convolutional Neural Network
Other Titles: Design and Development of Transformer Oil Quality Detection Tools using the Convolutional Neural Network Method
Authors: Valentino, Andi Philip
metadata.dc.contributor.advisor: Satria, Habib
Keywords: minyak transformator;convolutional neural network;matlab;warna;sensor;transformer oil;convolutional neural network;matlab;color;sensor
Issue Date: 19-Sep-2024
Publisher: UNIVERSITAS MEDAN AREA
Series/Report no.: NPM;208120023
Abstract: Kualitas minyak transformator merupakan faktor yang sangat mempengaruhi kinerja dan umur transformator. Penurunan kualitas minyak dapat menyebabkan kerusakan dan kegagalan operasi transformator. Tujuan penulisan skripsi ini adalah untuk merancang dan mengembangkan alat yang dapat mendeteksi kualitas minyak transformator dengan menggunakan metode Convolutional Neural Network (CNN) di MATLAB. Alat ini mengintegrasikan sensor optik untuk mendeteksi perubahan warna minyak transformator, yang kemudian dianalisis menggunakan model CNN untuk mengidentifikasi dan mengklasifikasikan kualitas minyak berdasarkan warna. Proses pengembangan melibatkan pengumpulan data gambar minyak transformator dengan berbagai tingkat kualitas, pelatihan model CNN, serta pengujian dan optimasi model. Hasil penelitian menunjukkan bahwa metode CNN yang diimplementasikan di MATLAB mampu memberikan akurasi tinggi dalam pendeteksian kualitas minyak transformator berdasarkan analisis warna, dengan tingkat kesalahan yang rendah. Tingkatan warna yang digunakan adalah skala warna ASTM D-1500. Terdapat empat sampel minyak yang akan diukur kualitasnya melalui warnanya. Sampel minyak pertama dan kedua menunjukkan hasil warna berada di rentang 0.5 – 2.0, yang artinya kualitas minyak masih bagus. Sampel minyak ketiga menunjukkan warna berada di rentang 2.5 – 4.0, yang menandakan kualitas minyak normal, dan sampel minyak keempat berada di rentang 4.5 – 6.0, yang menandakan kualitas minyak buruk. Diharapkan alat ini akan memberikan solusi yang efisien dan akurat untuk memeriksa kondisi minyak transformator, sehingga dapat mencegah kerusakan, meningkatkan reliabilitas operasional, dan mengurangi biaya perawatan. The quality of transformer oil is a factor that greatly influences the performance and life of the transformer. Decreased oil quality can cause damage and failure of transformer operation. The aim of writing this thesis is to design and develop a tool that can detect transformer oil quality using the Convolutional Neural Network (CNN) method in MATLAB. This tool integrates an optical sensor to detect changes in transformer oil color, which is then analyzed using a CNN model to identify and classify oil quality based on color. The development process involves collecting transformer oil image data with varying levels of quality, training a CNN model, and testing and optimizing the model. The research results show that the CNN method implemented in MATLAB is able to provide high accuracy in detecting transformer oil quality based on color analysis, with a low error rate. The color levels used are the ASTM D-1500 color scale. There are four oil samples whose quality will be measured through their color. The first and second oil samples showed color results in the range 0.5 – 2.0, which means the quality of the oil is still good. The third oil sample showed a color in the range 2.5 – 4.0, which indicates normal oil quality, and the fourth oil sample was in the range 4.5 – 6.0, which indicates poor oil quality. It is hoped that this tool will provide an efficient and accurate solution for checking the condition of transformer oil, thereby preventing damage, increasing operational reliability and reducing maintenance costs.
Description: 27 Halaman
URI: https://repositori.uma.ac.id/handle/123456789/26177
Appears in Collections:SP - Electrical Engineering

Files in This Item:
File Description SizeFormat 
208120023 - Andi Philip Valentino - Fulltext.pdfCover, Abstract, Chapter I, II, III, V, Bibliography1.05 MBAdobe PDFView/Open
208120023 - Andi Philip Valentino - Chapter IV.pdf
  Restricted Access
Chapter IV464.26 kBAdobe PDFView/Open Request a copy


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.