Adaptive Load Shedding Berbasis Stabilitas Tegangan Pada Sistem Tambora Menggunakan Algoritma Xgboost

Wicaksono, Wahyudi (2025) Adaptive Load Shedding Berbasis Stabilitas Tegangan Pada Sistem Tambora Menggunakan Algoritma Xgboost. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6022231004_Wahyudi Wicaksono.pdf Restricted to Repository staff only Download (9MB) | Request a copy

Abstract

Meningkatnya kompleksitas sistem tenaga listrik modern menghadirkan tantangan signifikan dalam menjaga kestabilan frekuensi dan tegangan, khususnya saat terjadi gangguan besar seperti trip pembangkit. Penelitian ini mengusulkan skema Adaptive Load Shedding (ALS) berbasis machine learning menggunakan algoritma XGBoost untuk meningkatkan ketahanan sistem tenaga Tambora. Model klasifikasi XGBoost mampu mengidentifikasi tingkat keparahan gangguan secara real-time dan membagi respons sistem ke dalam tiga mode operasional: Tanpa Pelepasan Beban, Mode Eco, dan Mode Agresif. Alokasi beban yang akan dilepas dihitung menggunakan pendekatan Center of Inertia (COI) untuk membagi beban antara subsistem Bima dan Sumbawa secara proporsional. Selain itu, indeks Fast Voltage Stability Index (FVSI) digunakan untuk memprioritaskan pelepasan beban pada lokasi dengan risiko ketidakstabilan tegangan tertinggi. Secara kuantitatif, algoritma XGBoost menunjukkan kinerja terbaik dengan akurasi klasifikasi mencapai 99,03%, mengungguli model pembanding seperti Random Forest dan MLP Neural Network, serta memiliki waktu prediksi rendah sebesar 2,65 ms, menjadikannya ideal untuk aplikasi real-time. Hasil simulasi menunjukkan bahwa frekuensi steady-state sistem tetap berada dalam rentang 49,5–50,5 Hz, sementara seluruh tegangan bus terjaga dalam batas toleransi +5% hingga -10% terhadap tegangan nominal, sesuai standar Grid Code. Selain menjaga stabilitas sistem, pendekatan ini terbukti efektif dalam mengurangi risiko overshedding maupun undershedding, dengan melakukan estimasi kebutuhan pelepasan beban yang lebih presisi berdasarkan kondisi sistem secara dinamis. Skema ALS yang diusulkan terbukti mampu meningkatkan keandalan sistem dan mencegah kolaps tegangan maupun frekuensi akibat gangguan besar.
================================================================================================================================
The increasing complexity of modern power systems presents significant challenges in maintaining frequency and voltage stability, especially during major disturbances such as generator trips. This study proposes an Adaptive Load Shedding (ALS) scheme based on machine learning using the XGBoost algorithm to enhance the resilience of the Tambora power system. The XGBoost classifier enables real-time detection of disturbance severity and classifies system responses into three operational modes: No Load Shedding, Eco Mode, and Aggressive Mode. Load curtailment is proportionally allocated between the Bima and Sumbawa subsystems using the Center of Inertia (COI) approach. Additionally, the Fast Voltage Stability Index (FVSI) is employed to rank substations based on voltage vulnerability, allowing the system to prioritize load reduction at the most critical locations. Quantitatively, XGBoost demonstrated superior performance with a classification accuracy of 99.03%, outperforming benchmark models such as Random Forest and MLP Neural Network, while maintaining a low prediction time of 2.65 ms, making it suitable for real-time applications. Dynamic simulation results show that the system’s steady-state frequency remains within the acceptable range of 49.5–50.5 Hz, and all bus voltages stay within the standard tolerance of +5% to -10% of nominal voltage, in accordance with the Grid Code. Beyond maintaining system stability, the proposed approach effectively reduces the risk of both overshedding and undershedding by estimating load shedding requirements more accurately based on real-time system conditions. Overall, the ALS scheme proves to be a reliable and adaptive defense mechanism against major disturbances in power systems.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Pelepasan Beban Adaptif, Stabilitas Tegangan, Stabilitas Frekuensi, XGBoost, Sistem Tambora, Pusat Inersia, FVSI, Adaptive Load Shedding, Voltage Stability, Frequency Stability, XGBoost, Tambora System, Center of Inertia, FVSI
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1010 Electric power system stability. Electric filters, Passive.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	Wahyudi Wicaksono
Date Deposited:	25 Jul 2025 02:12
Last Modified:	25 Jul 2025 02:12
URI:	http://repository.its.ac.id/id/eprint/120878

Actions (login required)

View Item