Pengembangan Sistem Deteksi Dini Overheat Terminasi Kubikel 20 Kv Gardu Induk Menggunakan Recurrent Neural Network (RNN) dengan Input Data Thermal dan Arus

Tangkai, Amirul Malitang (2026) Pengembangan Sistem Deteksi Dini Overheat Terminasi Kubikel 20 Kv Gardu Induk Menggunakan Recurrent Neural Network (RNN) dengan Input Data Thermal dan Arus. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022241052-Master_Thesis.pdf Restricted to Repository staff only Download (16MB) | Request a copy

Abstract

Dalam sistem ketenagalistrikan, terminasi kabel pada kubikel outgoing 20 kV gardu induk memiliki peranan krusial sebagai titik awal penyaluran energi listrik ke jaringan distribusi. Kegagalan pada komponen ini, terutama akibat overheat, dapat menyebabkan gangguan serius seperti kebakaran, pemadaman tidak terjadwal, hingga kerugian energi yang tidak tersalurkan (energy not served). Oleh karena itu, diperlukan sistem pemantauan thermal yang tidak hanya mampu melakukan deteksi dini, tetapi juga prediksi potensi gangguan secara adaptif dan akurat. Penelitian ini mengembangkan sistem real time monitoring dan deteksi anomali thermal pada terminasi kubikel outgoing 20 kV menggunakan kombinasi data citra inframerah dari thermal imager dan data arus dari kWh meter yang terintegrasi dengan sistem SCADA PT PLN Batam. Citra inframerah dikonversi menjadi data numerik dan diproses menggunakan algoritma kecerdasan buatan Recurrent neural network (RNN) untuk mendeteksi dan memprediksi anomali thermal akibat abnormalitas peralatan seperti sambungan longgar, korosi terminal, isolasi rusak, atau beban berlebih. Hasil yang diharapkan dari penelitian ini adalah terciptanya sistem yang mampu mengenali pola perubahan temperatur secara adaptif terhadap kondisi variasi beban, mendeteksi potensi overheat sebelum melewati ambang batas pada sistem konvensional, serta mendukung pelaksanaan pemeliharaan berbasis kondisi (Condition based maintenance). Dengan demikian, sistem ini diharapkan dapat meningkatkan keandalan distribusi listrik, efisiensi pemeliharaan, dan umur peralatan di gardu induk PLN.
=================================================================================================================================
In the electrical power system, cable termination in the 20 kV outgoing cubicle of a substation plays a crucial role as the starting point for delivering electrical energy to the distribution network. Failures in this component, particularly due to overheating, can cause serious disruptions such as fires, unscheduled outages, and financial losses from energy not served (ENS). Therefore, a thermal monitoring system is required that is not only capable of early detection but is also adaptive and provides accurate prediction of potential faults. This research develops a real-time monitoring and thermal anomaly detection system for 20 kV outgoing cubicle terminations using a combination of infrared image data from a thermal imager and current data from a kWh meter integrated with PT PLN Batam's SCADA system. The infrared images are converted into numerical data and processed using a Recurrent Neural Network (RNN) artificial intelligence algorithm to detect and predict thermal anomalies caused by equipment abnormalities, such as loose connections, terminal corrosion, degraded insulation, or overloading. The expected outcome of this research is the creation of a system capable of adaptively recognizing temperature variation patterns under changing load conditions, detecting potential overheating before it exceeds conventional system thresholds, and supporting the implementation of condition-based maintenance (CBM). Ultimately, this system is expected to improve power distribution reliability, maintenance efficiency, and equipment lifespan at PLN substations.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	terminasi kubikel, thermal imager, deteksi anomali overheat, recurrent neural network, simscape MATLAB. cable termination, thermal imager, overheat anomaly detection, recurrent neural network, simscape MATLAB.
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK3030 Electric power distribution systems
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Amirul Malitang Tangkai
Date Deposited:	31 Jul 2026 07:04
Last Modified:	31 Jul 2026 07:04
URI:	http://repository.its.ac.id/id/eprint/140805

Actions (login required)

View Item