A Monitoring System for Online Fault Detection and Classification in Photovoltaic Plants

Area of Science:

• Renewable Energy Systems
• Electrical Engineering
• Artificial Intelligence

Background:

• Photovoltaic (PV) energy adoption is rising globally, driven by policies to reduce fossil fuel reliance.
• PV system efficiency is significantly impacted by environmental factors and operational faults, leading to energy losses.
• Effective monitoring and fault management are crucial for optimizing PV plant performance.

Purpose of the Study:

• To develop an integrated monitoring system (MS) for real-time and historical data collection of PV electrical and environmental variables.
• To propose a recursive linear model for detecting faults in PV systems using irradiance and temperature as inputs.
• To implement a machine-learning-based method for classifying detected PV faults.

Main Methods:

• Development of a Monitoring System (MS) to capture instantaneous and historical electrical and environmental data.
• Implementation of a recursive linear model for fault detection, utilizing PV panel irradiance and temperature to predict power output.
• Application of an Artificial Neural Network (ANN) model for classifying faults into categories such as short-circuit, open-circuit, partial shadowing, and degradation.

Main Results:

• The fault detection model achieved an accuracy of 93.09% for a 5 kW PV plant over 16 days with 143 hours of faults.
• The ANN-based fault classification model demonstrated an accuracy of 95.44% for the same dataset.
• The integrated system achieved an overall accuracy of 92.64% for combined fault detection and classification.

Conclusions:

• The developed integrated system offers online identification and classification of PV faults, a novel contribution compared to existing methods.
• The system provides real-time and historical monitoring of PV plant parameters, enhancing operational efficiency.
• This approach significantly improves the reliability and performance of photovoltaic energy systems.