Real-Time solar photovoltaic performance analysis and fault detection using a Random Forest algorithm
|
Full Text |
Pdf
|
|
Author |
Muhammad Afif Amran, Azmi Awang Md Isa, Muhammad Idzdihar Idris, Zahariah Manap, Mohd Shahril Md Zin, Sani Irwan Md Salim, Kamarul Ariffin Noordin and Mothana Lafta Attiah
|
|
e-ISSN |
1819-6608 |
|
On Pages
|
443-452
|
|
Volume No. |
21
|
|
Issue No. |
7
|
|
Issue Date |
June 10, 2026
|
|
DOI |
https://doi.org/10.59018/042652
|
|
Keywords |
solar photovoltaic monitoring, internet of things (IoT), random forest algorithm, real-time fault detection.
|
Abstract
The increasing adoption of solar photovoltaic (PV) systems highlights the need for intelligent monitoring to maintain efficiency and reliability. Conventional monitoring methods are often passive, lacking real-time analytics and automated fault detection, which can lead to undetected performance issues and higher maintenance costs. This paper develops an IoT-based solar energy monitoring system with a web interface for real-time performance evaluation and predictive fault detection. The system continuously measures voltage, current, power, and temperature using an Arduino Uno and transmits data via an ESP32 to the ThingSpeak cloud. A Random Forest model, trained in Google Colab and deployed on a Raspberry Pi through a Python Flask API, enables real-time classification of normal operation, partial shading, open circuit, and overheating faults. Data and fault status are visualized on a custom HTML dashboard with automated alert notifications. Experimental results demonstrate that the integrated system achieves an overall classification accuracy of 100%, effectively distinguishing between four distinct operational states: Normal, Partial Shading, Open Circuit, and Overheating under varying environmental conditions. The validation also confirms a low-latency alert mechanism, with Telegram notifications delivered within 1-3 seconds of fault detection. The system enhances PV reliability and efficiency by enabling proactive maintenance, minimizing downtime, and supporting sustainable energy management through the integration of IoT hardware and machine learning.
Back