View in EDS

Visual fault detection in photovoltaic modules using decision tree algorithms with deep learning features.

Saved in:
Bibliographic Details
Title: Visual fault detection in photovoltaic modules using decision tree algorithms with deep learning features.
Authors: Sridharan, Naveen Venkatesh, Sugumaran, Vaithiyanathan
Source: Energy Sources Part A: Recovery, Utilization & Environmental Effects; Dec2025, Vol. 47 Issue 2, p1-17, 17p
Subject Terms: PHOTOVOLTAIC power generation, DECISION trees, DEEP learning, DRONE aircraft equipment & supplies, CLASSIFICATION, RANDOM forest algorithms, IMAGE recognition (Computer vision), INSPECTION & review
Abstract: Visual faults in photovoltaic (PV) modules persist as a problem that can create consequences such as reduced life span, increased output power loss and raising safety concerns during operation. Increased manpower requirement, larger time consumption, confinement to single fault prediction and high initial cost are certain drawbacks of conventional fault diagnosis techniques. Recent advancements in technology and the innovation of scientific techniques have urged the need for advanced fault diagnosis techniques that deliver instantaneous results. In the present study, unmanned aerial vehicles (UAVs) were employed to capture images of PVM with visual faults. The most common visual faults in photovoltaic modules (PVM) are delamination, burn marks, glass breakage, discoloration, and snail trails. Each fault condition contains a unique image pattern appearance attributed to the particular type of fault. Such patterns are extracted using convolutional neural networks and classified with the help of decision tree algorithms. First, the features are extracted from these aerial images by utilizing pre-trained AlexNet convolutional neural networks. Secondly, the J48 decision tree algorithm is utilized to select the most significant and valuable features from the extracted image features. Finally, the classification is carried out with several decision tree algorithms such as decision stump, hoeffiding tree, J48, linear model tree (LMT), random forest, random tree, representative (REP) tree, best first (BF) tree, extra tree, functional tree (FT), J48 consolidated, J48 graft, least absolute deviation (LAD) tree, naïve bayes (NB) tree and simple cart. The classification accuracies of the algorithms mentioned above are compared to suggest the best-in-class algorithm for real-time application. Among all the available tree-based algorithms, the random forest algorithm produced a maximum classification accuracy of 98.25% with a computational time of 0.89 seconds. [ABSTRACT FROM AUTHOR]
Copyright of Energy Sources Part A: Recovery, Utilization & Environmental Effects is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Complementary Index
Description
Abstract:Visual faults in photovoltaic (PV) modules persist as a problem that can create consequences such as reduced life span, increased output power loss and raising safety concerns during operation. Increased manpower requirement, larger time consumption, confinement to single fault prediction and high initial cost are certain drawbacks of conventional fault diagnosis techniques. Recent advancements in technology and the innovation of scientific techniques have urged the need for advanced fault diagnosis techniques that deliver instantaneous results. In the present study, unmanned aerial vehicles (UAVs) were employed to capture images of PVM with visual faults. The most common visual faults in photovoltaic modules (PVM) are delamination, burn marks, glass breakage, discoloration, and snail trails. Each fault condition contains a unique image pattern appearance attributed to the particular type of fault. Such patterns are extracted using convolutional neural networks and classified with the help of decision tree algorithms. First, the features are extracted from these aerial images by utilizing pre-trained AlexNet convolutional neural networks. Secondly, the J48 decision tree algorithm is utilized to select the most significant and valuable features from the extracted image features. Finally, the classification is carried out with several decision tree algorithms such as decision stump, hoeffiding tree, J48, linear model tree (LMT), random forest, random tree, representative (REP) tree, best first (BF) tree, extra tree, functional tree (FT), J48 consolidated, J48 graft, least absolute deviation (LAD) tree, naïve bayes (NB) tree and simple cart. The classification accuracies of the algorithms mentioned above are compared to suggest the best-in-class algorithm for real-time application. Among all the available tree-based algorithms, the random forest algorithm produced a maximum classification accuracy of 98.25% with a computational time of 0.89 seconds. [ABSTRACT FROM AUTHOR]
ISSN:15567036
DOI:10.1080/15567036.2021.2020379