Deteksi Arc Fault Seri Berbasis Hilbert-Huang Transform (HHT) Dan Convolutional Neural Network (CNN) Pada Sistem AC Tegangan Rendah

Faizin, Alfa Kusnal (2025) Deteksi Arc Fault Seri Berbasis Hilbert-Huang Transform (HHT) Dan Convolutional Neural Network (CNN) Pada Sistem AC Tegangan Rendah. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022231056_Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy

Abstract

Penelitian ini membahas deteksi gangguan busur listrik (arc fault) seri pada sistem AC tegangan rendah menggunakan Hilbert-Huang Transform (HHT) dan Convolutional Neural Network (CNN). Fokus penelitian adalah mengklasifikasikan kondisi arc fault yang terdampak oleh variasi gangguan tegangan atau Voltage Variation Disturbances (VVD), seperti harmonik, voltage sag, dan voltage swell. Empirical Mode Decomposition (EMD) diterapkan pada sinyal arus untuk memperoleh Intrinsic Mode Function (IMF), di mana IMF ke-1 hingga ke-4 dikombinasikan dan dianalisis menggunakan Transformasi Hilbert untuk menghasilkan spektrum waktu-frekuensi. Spektrum ini divisualisasikan dalam bentuk citra heatmap resolusi 224×224 piksel dan digunakan sebagai input CNN. Lima kondisi diuji: kondisi normal, arc fault murni, arc fault dengan harmonik, arc fault dengan voltage sag, dan arc fault dengan voltage swell. Dataset berisi 750 citra, diperoleh dari simulasi MATLAB dan data eksperimen laboratorium. Pelatihan CNN dilakukan menggunakan arsitektur GoogLeNet di MATLAB, dengan rasio data pelatihan:validasi:pengujian sebesar 80%:10%:10%. Hasil menunjukkan bahwa CNN mampu mengklasifikasikan kondisi arc fault yang terpengaruh VVD dengan akurasi validasi mencapai 93,33% dan nilai loss di bawah 0.01. Confusion matrix menunjukkan akurasi klasifikasi per kelas di atas 95% kecuali pada Voltage Sag. Penelitian ini membuktikan efektivitas kombinasi HHT dan CNN dalam mendeteksi dan membedakan gangguan arc fault seri secara akurat.
==================================================================================================================================
This study addresses the detection of series arc faults in low-voltage AC systems using the Hilbert-Huang Transform (HHT) and Convolutional Neural Network (CNN). The focus is on classifying arc fault conditions affected by various Voltage Variation Disturbances (VVD), such as harmonics, voltage sag, and voltage swell. Empirical Mode Decomposition (EMD) is applied to the current signal to extract Intrinsic Mode Functions (IMFs), where the first to fourth IMFs are combined and analyzed using the Hilbert Transform to produce a time- frequency spectrum. This spectrum is visualized as a heatmap image with a resolution of 224×224 pixels and used as input to the CNN. Five conditions are tested: normal, pure arc fault, arc fault with harmonics, arc fault with voltage sag, and arc fault with voltage swell. A dataset of 750 images was generated from MATLAB simulations and laboratory experiments. CNN training was performed using the GoogLeNet architecture in MATLAB, with a training:validation:test split of 80%:10%:10%. The results show that the CNN can classify arc fault conditions influenced by VVD with a validation accuracy of 93,3% and a loss value below 0.01. The confusion matrix indicates that the classification accuracy for each class is above 95%, except for the Voltage Sag condition. This study demonstrates the effectiveness of combining HHT and CNN in accurately detecting and distinguishing series arc faults.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Arc fault, Gangguan Tegangan (VVD), Transformasi Hilbert-Huang, CNN, Deteksi Gangguan. Arc fault, VVD, Hilbert-Huang Transform, CNN, fault detection.
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA336 Artificial Intelligence T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Alfa Kusnal Faizin
Date Deposited:	24 Jul 2025 07:21
Last Modified:	24 Jul 2025 07:22
URI:	http://repository.its.ac.id/id/eprint/121049

Actions (login required)

View Item