Nugroho, Khoiru Adi and -, Hasyim Asy’ari, S.T.,M.T. (2026) OPTIMASI POSISI PANEL SURYA DUAL-AXIS MENGGUNAKAN PEN-DEKATAN MACHINE LEARNING UNTUK MEMAKSIMALKAN DAYA KELUARAN. Skripsi thesis, Universitas Muhammadiyah Surakarta.
![[img]](https://eprints.ums.ac.id/style/images/fileicons/application_pdf.png) |
PDF (Naskah Publikasi)
NASPUB_KHOIRU ADI N_D400220042 (1).pdf Download (1MB) |
|
PDF (Surat Pernyataan Publikasi)
pernyataan .pdf Restricted to Repository staff only Download (447kB) |
Abstract
This study aims to design an Internet of Things (IoT)-based dual-axis solar panel optimization system using a machine learning approach to maximize solar panel output power. The main equipment used includes a GH Solar 30 Wp Solar Panel, a 20A Solar Charge Controller (SCC), an Inforce 12V/9AH Battery, servo motors as panel actuators, a BH1750 sensor to measure light intensity (lux), an INA219 sensor to measure voltage, current, and power, a DS18B20 sensor to measure panel temperature, and an RTC module as a time reference for data acquisition. Data were collected every 10 seconds from 09:00 AM to 03:00 PM, resulting in approximately 1676 measurement data points. All data were processed using ESP32 and displayed in real-time through an IoT platform. The XGBoost algorithm was trained first due to its high accuracy, and its prediction results were then used as a reference to train a simpler Decision Tree model that could be implemented on Arduino IDE. Based on the testing results before machine learning implementa-tion, the average light intensity was 30,259 lux, voltage was 15.54 V, current was 1.15 A, output power was 17.95 W, and panel temperature was 45.97°C. After machine learning implementation, the average light intensity became 25,175 lux, voltage was 16.21 V, current was 0.99 A, output power was 16.34 W, and panel temperature was 44.62°C. The analysis results showed that the power-to-light intensity ratio increased from 0.000593 W/lux to 0.000649 W/lux, or improved by approximately 13.37%, indicating that the system could generate more optimal power even under lower light intensity conditions. The 3.15% reduction in panel temperature also indicates that the machine learning-based dual-axis system was able to reduce thermal losses through more optimal panel positioning.
| Item Type: | Thesis (Skripsi) |
|---|---|
| Uncontrolled Keywords: | Keywords: Solar Panel, Dual-Axis Tracker, Internet of Things, Machine Learning, XGBoost, Decision Tree, Power Optimization. |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK01 Weak flow T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK02 High Voltage |
| Divisions: | Fakultas Teknik > S1 Teknik Elektro |
| Depositing User: | KHOIRU ADI NUGROHO |
| Date Deposited: | 14 Aug 2026 04:20 |
| Last Modified: | 14 Aug 2026 04:20 |
| URI: | http://eprints.ums.ac.id/id/eprint/147761 |
Actions (login required)
 |
View Item |
