Convolutional Neural Network for Dust and Hotspot Classification in PV Modules

Dual-input robust diagnostics for railway point machines via audio signals

Railway Point Machine (RPM) is a fundamental component of railway infrastructure and plays a crucial role in ensuring the safe operation of trains. Its primary function is to divert trains from one track to another, enabling connections between different lines and facilitating route selection. By judiciously deploying turnouts, railway systems can provide efficient transportation services while ensuring the safety of passengers and cargo. As signal processing technologies develop rapidly, taking the easy acquisition advantages of audio signals, a fault diagnosis method for RPMs is proposed by considering noise and multi-channel signals. The proposed method consists of several stages. Initially, the signal is subjected to pre-processing steps, including cropping and channel separation. Subsequently, the signal undergoes noise addition using the Random Length and Dynamic Position Noises Superposition (RDS) module, followed by conversion to a greyscale image. To enhance the data, Synthetic Minority Oversampling Technique (SMOTE) module is applied. Finally, the training data is fed into a Dual-input Attention Convolutional Neural Network (DIACNN). By employing various experimental techniques and designing diverse datasets, our proposed method demonstrates excellent robustness and achieves an outstanding classification accuracy of 99.73%.

Detection and analysis of deteriorated areas in solar PV modules using unsupervised sensing algorithms and 3D augmented reality

Solar Photovoltaic (PV) systems are increasingly vital for enhancing energy security worldwide. However, their efficiency and power output can be significantly reduced by hotspots and snail trails, predominantly caused by cracks in PV modules. This article introduces a novel methodology for the automatic segmentation and analysis of such anomalies, utilizing unsupervised sensing algorithms coupled with 3D Augmented Reality (AR) for enhanced visualization. The methodology outperforms existing segmentation techniques, including Weka and the Meta Segment Anything Model (SAM), as demonstrated through computer simulations. These simulations were conducted using the Cali-Thermal Solar Panels and Solar Panel Infrared Image Datasets, with evaluation metrics such as the Jaccard Index, Dice Coefficient, Precision, and Recall, achieving scores of 0.76, 0.82, 0.90, 0.99, and 0.76, respectively. By integrating drone technology, the proposed approach aims to revolutionize PV maintenance by facilitating real-time, automated solar panel detection. This advancement promises substantial cost reductions, heightened energy production, and improved performance of solar PV installations. Furthermore, the innovative integration of unsupervised sensing algorithms with 3D AR visualization opens new avenues for future research and development in the field of solar PV maintenance.

Photovoltaics Plant Fault Detection Using Deep Learning Techniques

Solar energy is the fastest-growing clean and sustainable energy source, outperforming other forms of energy generation. Usually, solar panels are low maintenance and do not require permanent service. However, plenty of problems can result in a production loss of up to ~20% since a failed panel will impact the generation of a whole array. High-quality and timely maintenance of the power plant will reduce the cost of its repair and, most importantly, increase the life of the power plant and the total generation of electricity. Manual monitoring of panels is costly and time-consuming on large solar plantations; moreover, solar plantations located distantly are more complicated for humans to access. This paper presents deep learning-based photovoltaics fault detection techniques using thermal images obtained from an unmanned aerial vehicle (UAV) equipped with infrared sensors. We implemented the three most accurate segmentation models to detect defective panels on large solar plantations. The models employed in this work are DeepLabV3+, Feature Pyramid Network (FPN) and U-Net with different encoder architectures. The obtained results revealed intersection over union (IoU) of 79%, 85%, 86%, and dice coefficients of 87%, 92%, 94% for DeepLabV3+, FPN, and U-Net, respectively. The implemented models showed efficient performance and proved effective to resolve these challenges.

Automatic Inspection of Photovoltaic Power Plants Using Aerial Infrared Thermography: A Review

In recent years, aerial infrared thermography (aIRT), as a cost-efficient inspection method, has been demonstrated to be a reliable technique for failure detection in photovoltaic (PV) systems. This method aims to quickly perform a comprehensive monitoring of PV power plants, from the commissioning phase through its entire operational lifetime. This paper provides a review of reported methods in the literature for automating different tasks of the aIRT framework for PV system inspection. The related studies were reviewed for digital image processing (DIP), classification and deep learning techniques. Most of these studies were focused on autonomous fault detection and classification of PV plants using visual, IRT and aIRT images with accuracies up to 90%. On the other hand, only a few studies explored the automation of other parts of the procedure of aIRT, such as the optimal path planning, the orthomosaicking of the acquired images and the detection of soiling over the modules. Algorithms for the detection and segmentation of PV modules achieved a maximum F1 score (harmonic mean of precision and recall) of 98.4%. The accuracy, robustness and generalization of the developed algorithms are still the main issues of these studies, especially when dealing with more classes of faults and the inspection of large-scale PV plants. Therefore, the autonomous procedure and classification task must still be explored to enhance the performance and applicability of the aIRT method.

Photovoltaic Panels Classification Using Isolated and Transfer Learned Deep Neural Models Using Infrared Thermographic Images

Defective PV panels reduce the efficiency of the whole PV string, causing loss of investment by decreasing its efficiency and lifetime. In this study, firstly, an isolated convolution neural model (ICNM) was prepared from scratch to classify the infrared images of PV panels based on their health, i.e., healthy, hotspot, and faulty. The ICNM occupies the least memory, and it also has the simplest architecture, lowest execution time, and an accuracy of 96% compared to transfer learned pre-trained ShuffleNet, GoogleNet, and SqueezeNet models. Afterward, ICNM, based on its advantages, is reused through transfer learning to classify the defects of PV panels into five classes, i.e., bird drop, single, patchwork, horizontally aligned string, and block with 97.62% testing accuracy. This proposed approach can identify and classify the PV panels based on their health and defects faster with high accuracy and occupies the least amount of the system's memory, resulting in savings in the PV investment.