Analysis and techniques of partial shading detection and classification in solar photovoltaic arrays

Partial shading in solar photovoltaic (PV) modules typically reduces the output current of the shaded PV module due to the reduction in the irradiance level. Although this phenomenon is temporary in nature, it is considered an intermittent fault, and it is essential for a protection system to differentiate it from other fault conditions to avoid unnecessary tripping. The main problem in identifying partial shading in a PV system is the difficulty of extracting its features under different shading conditions. To address this difficulty, this study proposes a novel approach combining Wavelet Packet Transform (WPT) along with Empirical Mode Decomposition (EMD) to extract the features of PV panel output voltage and string current signals during partial shading conditions. In the first stage, the WPT is used to split the PV voltage and string currents into specific sub-band frequencies, and then EMD is used to decompose the selected frequency bands into a number of intrinsic mode functions (IMFs) and a residual. The generated IMF components are then fed into different machine-learning models. The results highlight the Random Forest (RF) model's potential as a reliable technique for diagnosing partial shading in PV systems, ensuring that shading faults are detected and addressed promptly. In contrast, while the K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Artificial Neural Networks (ANN), Gradient Boosting, and Adaptive Boosting (AdaBoost) models also demonstrated acceptable performance in detecting partial shading, they were less effective in classifying partial shading conditions. This proposed hybrid technique provides a high-resolution representation of the array voltage and string currents without loss in the time-frequency resolution, aiding in the detection of partial shading and differentiation of its strength. The results indicate that the proposed approach achieved a detection accuracy of 98.4% and a classification accuracy of 97.6%.

Name thesis:degree_name

Doctor of Philosophy (PhD)

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2025

Author dc:creator

• Abdulmawjood, Kais A.N. Ridha

Advisor dc:contributor.advisor

• Ibrahim , Walid Morsi

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/1979

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/1979