Photovoltaic Array Fault Diagnosis and Localization Method Based on Modulated Photocurrent and Machine Learning.

Area of Science:

• Renewable Energy Systems
• Electrical Engineering
• Materials Science

Background:

• Photovoltaic (PV) arrays degrade over time due to outdoor exposure, leading to various faults.
• Effective fault diagnosis systems are crucial for PV array reliability and performance.
• Current methods often compromise between diagnostic accuracy and fault localization capabilities.

Purpose of the Study:

• To develop a fault identification and localization approach for PV arrays.
• To overcome the limitations of existing methods by achieving both high accuracy and precise localization.
• To enable rapid and cost-effective PV array fault detection.

Main Methods:

• Utilizing modulated photocurrent and machine learning for fault diagnosis.
• Employing frequency-modulated light to separate photocurrent and measure individual panel efficiency.
• Applying machine learning classification algorithms to analyze current amplitude and frequency for fault identification.

Main Results:

• Achieved high-speed (5800 obs/s) and high-accuracy (97.8%) fault identification and localization using a neural network algorithm.
• Demonstrated the method's effectiveness through practical experimentation.
• Confirmed the ability to identify faults by measuring only the short-circuit current.

Conclusions:

• The proposed modulated photocurrent and machine learning method offers a superior solution for PV array fault diagnosis.
• This approach provides a practical, low-cost, high-speed, and highly accurate system for fault identification and localization.
• The findings support the widespread adoption of this technique for enhancing PV system reliability.