Sistem Rekomendasi Penentuan Jenis Pembangkit Listrik Berbasis Data Cuaca Lokal Menggunakan Algoritma Random Forest

Ramadhani, Sabrina (2026) Sistem Rekomendasi Penentuan Jenis Pembangkit Listrik Berbasis Data Cuaca Lokal Menggunakan Algoritma Random Forest. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221043-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Pemilihan jenis pembangkit energi terbarukan memerlukan analisis kondisi cuaca yang sesuai dengan karakteristik lokasi. Penelitian ini mengembangkan sistem rekomendasi awal untuk menentukan jenis Pembangkit Listrik Tenaga Surya (PLTS), Pembangkit Listrik Tenaga Bayu (PLTB), Hybrid, atau Tidak Direkomendasikan menggunakan algoritma Random Forest. Dataset diperoleh dari Open-Meteo pada delapan lokasi dengan tujuh parameter cuaca, yaitu radiasi matahari, kecepatan angin, curah hujan, temperatur, kelembapan, tekanan udara, dan arah angin. Data tahun 2024 sebanyak 2.928 data harian digunakan sebagai data training dan diseimbangkan menggunakan Random Under-Sampling menjadi 251 data per kelas atau 1.004 data. Data tahun 2025 digunakan sebagai data testing balanced sebanyak 296 data per kelas atau 1.184 data. Hasil training Random Forest memperoleh accuracy sebesar 94,8207%, sedangkan hasil testing memperoleh accuracy sebesar 64,3581%. Hasil tersebut lebih tinggi dibandingkan Decision Tree yang memperoleh accuracy testing sebesar 58,6993%. Pengujian sistem menggunakan 8.640 data weather station Ciputri yang diagregasikan menjadi 30 data harian menghasilkan accuracy sebesar 70,0000%, precision kelas PLTS sebesar 100,0000%, recallsebesar 70,0000%, dan F1-score sebesar 82,3529%. Hasil pengujian memberikan rekomendasi utama berupa PLTS dengan energi spesifik sebesar 4,20 kWh/kWp/hari. Perhitungan potensi energi angin berdasarkan interval lima menit menghasilkan rata-rata sebesar 1,35 kWh/m²/hari, dengan durasi angin efektif sekitar 16,28 jam per hari dan tingkat ketersediaan sebesar 67,85%. Durasi angin efektif digunakan untuk menunjukkan lama dan frekuensi terjadinya kondisi angin yang memenuhi threshold selama periode pengamatan. Sistem iimplementasikan dalam dashboard lokal yang menampilkan rekomendasi, probabilitas kelas, serta estimasi potensi energi sebagai pendukung feasibility study awal.
=====================================================================================================================================
Selecting an appropriate type of renewable energy power plant requires an analysis of weather conditions that reflect the characteristics of the location. This study develops an initial recommendation system to determine the suitability of Solar Power Plants, Wind Power Plant, Hybrid systems, or Not Recommended conditions using the Random Forest algorithm. The dataset was obtained from Open-Meteo for eight locations using seven weather parameters, namely solar radiation, wind speed, rainfall, temperature, humidity, air pressure, and wind direction. A total of 2,928 daily data records from 2024 were used as training data and balanced using Random Under-Sampling into 251 records per class, resulting in 1,004 records. The 2025 dataset was used as balanced testing data, consisting of 296 records per class or 1,184 records in total. The Random Forest model achieved a training accuracy of 94.8207% and a testing accuracy of 64.3581%. This result was higher than the testing accuracy of the Decision Tree model, which reached 58.6993%. System testing using 8,640 records from the Ciputri weather station, aggregated into 30 daily records, achieved an accuracy of 70.0000%, a precision of 100.0000% for the PLTS class, a recall of 70.0000%, and an F1-score of 82.3529%. The testing results provided PLTS as the primary recommendation, with a specific energy potential of 4.20 kWh/kWp/day. The wind energy potential calculated at five-minute intervals produced an average of 1.35 kWh/m²/day, with an effective wind duration of approximately 16.28 hours per day and an availability rate of 67.85%. The effective wind duration was used to indicate the duration and frequency of wind conditions that met the threshold during the observation period. The system was implemented through a local dashboard that displays the recommendation, class probabilities, and estimated energy potential to support an initial feasibility study.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Data Cuaca Lokal, Energi Terbarukan, PLTS, PLTB, Random Forest, Sistem Rekomendasi. Local Weather Data, Renewabl
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1001 Production of electric energy or power T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1087 Photovoltaic power generation T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK1322.6 Electric power-plants T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105.546 Computer algorithms T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.674 Detectors. Sensors
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Sabrina Ramadhani
Date Deposited:	10 Aug 2026 00:46
Last Modified:	10 Aug 2026 00:46
URI:	http://repository.its.ac.id/id/eprint/144233

Actions (login required)

View Item