Open Collections

Martinez, Gervasio

University of British Columbia

Photovoltaic (PV) systems are an increasingly important source of renewable energy, providing clean, sustainable power for residential and industrial applications. However, they face safety challenges, particularly the risk of arc faults, which occur when electrical current flows through unintended paths, potentially leading to fires or safety risks. Servicing the installation helps prevent this issue, but it is typically expensive and seldom carried out. Therefore, detecting arc faults becomes crucial not only to prevent these hazards but also to comply with safety standards such as UL1699B. This work explores several arc fault detection methods, focusing on computationally efficient techniques suitable for low-cost microcontrollers commonly found in commercial PV systems. Key techniques examined include frequency-domain analysis, wavelet transforms, and digital signal processing, all chosen to balance performance with the hardware limitations imposed by cost constraints. By expanding on the standard test cases defined in UL1699B, a more comprehensive and robust analysis was performed, introducing various real-world disturbances such as converter noise and power line communication (PLC) signals. These additional perturbations allow for more rigorous testing of the detection methods, ensuring a robust performance under challenging operational conditions and reducing false positives. The proposed WIK strategy, which integrates Wavelet transforms, IIR filtering, and Kurtosis analysis, is introduced as an innovative and highly effective solution for arc fault detection. This method balances high detection accuracy with low computational demand, making it ideal for cost-sensitive applications. Extensive testing shows that the WIK strategy is not only compliant with safety standards but also capable of filtering out noise and disturbances, ensuring optimal performance, reliability, and safety for PV installations.

Text

2024-10-10

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/89381

Electrical and Computer Engineering

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nc-nd/4.0/