Shading fault detection in a grid-connected PV system using vertices principal component analysis
Document Type
Article
Publication Date
2-1-2021
Abstract
Partial shading severely impacts the performance of the photovoltaic (PV) system by causing power losses and creating hotspots across the shaded cells or modules. Proper detection of shading faults serves not only in harvesting the desired power from the PV system, which helps to make solar power a reliable renewable source, but also helps promote solar versus other fossil fuel electricity-generation options that prevent making climate change targets (e.g. 2015's Paris Agreement) achievable. This work focuses primarily on detecting partial shading faults using the vertices principal component analysis (VPCA), a data-driven method that combines the simplicity of its linear model and the ability to consider the uncertainties of the different measurements of a PV system in an interval format. Data from a grid connected monocrystalline PV array, installed on the rooftop of the Power Electronics and Renewable Energy Research Laboratory (PEARL), University of Malaya, Malaysia, have been used to train the VPCA model. To prove the effectiveness of this VPCA method, four partial shading patterns have been created. The obtained performance has, then, been tested against a regular PCA. In addition to its ability to acknowledge the uncertainty of a PV system, the VPCA method has shown an enhanced performance of detecting partial shading fault in comparison with the standard PCA. Also, included in the article is an extension of the contribution plot diagnosis-based method, of the Q-statistic, to the interval-valued case aiming to pinpoint the out-of-control variables. (c) 2020 Published by Elsevier Ltd.
Keywords
Photovoltaic system (PV), Partial shading, Fault detection, Fault diagnosis, Principal component analysis (PCA), Interval-valued PCA
Divisions
fac_eng
Funders
la Direction Generale de la Recherche Scientifique et du Developpement Technologique, Algeria (DGRSDT) (A01L08UN350120200002)
Publication Title
Renewable Energy
Volume
164
Publisher
Elsevier
Publisher Location
THE BOULEVARD, LANGFORD LANE, KIDLINGTON, OXFORD OX5 1GB, ENGLAND