Bogdan Marian Diaconu

Convolutional neural networks (CNNs) can achieve high accuracy in photovoltaic (PV) fault recognition from RGB imagery, yet their decisions may rely on shortcut cues induced by heterogeneous backgrounds, viewpoints, and class imbalance. This work presents a multi-method explainability audit on the Kaggle PV Panel Defect Dataset (six classes), comparing five architectures (Baseline CNN, VGG16, ResNet50, InceptionV3, EfficientNetB0). Explanations are obtained with LIME superpixel surrogates (reported together with kernel-weighted surrogate fidelity), occlusion sensitivity (quantified via IoU@Top10% against consistent proxy masks, Shannon entropy, and Hoyer sparsity), and Integrated Gradients evaluated by deletion–insertion faithfulness and a Faithfulness Gap. While ResNet50 yields the best predictive performance, EfficientNetB0 shows the most consistent faithfulness evidence and stable panel-centered attributions. The analysis highlights class-dependent vulnerability to context cues, especially for the Clean and damaged classes, and supports using quantitative explainability diagnostics during model selection and dataset curation to mitigate shortcuts in vision-based PV monitoring.

Convolutional neural networks are increasingly used for photovoltaic fault recognition from RGB imagery, yet high benchmark accuracy can mask shortcut learning induced by heterogeneous backgrounds, viewpoints and class imbalance. Using the Kaggle "PV Panel Defect Dataset" dataset, we compare five architectures (Baseline CNN, VGG16, ResNet50, InceptionV3 and EfficientNetB0) through a complementary explainability pipeline: LIME superpixel surrogates (with kernel-weighted R2 fidelity), occlusion sensitivity (functional relevance under localized masking) and Integrated Gradients (IG) validated by deletion-insertion curves. To reduce reliance on subjective saliency inspection, we quantify localization and concentration using IoU@Top10% against consistent proxy defect masks, Shannon entropy and Hoyer sparsity, and we summarize IG faithfulness with a Faithfulness Gap (AUC_insertion - AUC_deletion) and an accuracy-faithfulness consistency score at class level. ResNet50 attains the best predictive performance (82.3% accuracy), while EfficientNetB0 provides the strongest overall evidence faithfulness (mean Faithfulness Gap ~ 0.019) and stable, panel-centered attributions. InceptionV3 frequently yields diffuse relevance, and VGG16 produces highly concentrated but occasionally brittle hotspots. Bird-drop and Snow-covered show the most consistent alignment between accuracy and faithful evidence, whereas Clean and the two damage classes remain vulnerable to context cues (e.g., borders and background textures). The results support integrating quantitative explainability diagnostics into PV model selection and dataset curation to mitigate shortcuts and improve trustworthiness in vision-based PV monitoring.

Electrical cable insulation, mainly composed of polymeric materials, progressively deteriorates under thermal, electrical, mechanical, and environmental stress factors. This degradation reduces dielectric strength, thermal stability, and mechanical integrity, thereby increasing susceptibility to failure modes such as partial discharges, arcing, and surface tracking—recognized precursors of fire ignition. This review consolidates current knowledge on the degradation pathways of cable insulation and their direct link to fire hazards. Emphasis is placed on mechanisms including thermal-oxidative aging, electrical treeing, surface tracking, and thermal conductivity decline, as well as the complex interactions introduced by flame-retardant additives. A bibliometric analysis of 217 publications reveals strong clustering around material degradation phenomena, while underlining underexplored areas such as ignition mechanisms, diagnostic monitoring, and system-level fire modeling. Comparative experimental findings further demonstrate how insulation aging modifies ignition thresholds, heat release rates, and smoke toxicity. By integrating perspectives from materials science, electrical engineering, and fire dynamics, this review establishes the nexus between aging mechanisms and fire hazards.

Coal spontaneous combustion (CSC) is a multifaceted research domain that has been widely explored in the literature, ranging from analytical and numerical modeling to the development of fire suppression materials and methods. A comprehensive review of the literature has revealed several distinct research trajectories, or “roadmaps”, identified through criteria such as the volume of studies addressing each theme, the presence of review papers dedicated to a specific roadmap, and the explicit mention of coal spontaneous combustion in the title or keywords. This classification framework has outlined six primary roadmaps: (1) spread, quantification, and impact; (2) mechanisms, models, factors, and parameters; (3) experimental studies and models; (4) detection, monitoring, and prediction; (5) prevention and control; and (6) applications. While interconnections exist between these roadmaps, and all ultimately converge towards roadmap 5 (prevention and control), each roadmap constitutes a distinct research cluster. The focus of this review is on roadmap 4, specifically addressing the methods and technologies for detection, monitoring, and prediction of CSC events. This review encompasses studies published from 2010 to the present, providing a thorough examination of the various detection techniques employed, with particular emphasis on their limitations and the strategies proposed to overcome these challenges. A critical analysis highlights the key advantages and disadvantages of each category of techniques, offering insights into their practical applications and the potential for future advancements in this field. The present review aims to contribute to the refinement of detection and monitoring methods for CSC, with the goal of enhancing early detection capabilities and improving fire management strategies.

Fire detection is a critical safety issue due to the major and irreversible consequences of fire, from economic prejudices to loss of life. It is therefore of utmost importance to design reliable, automated systems that can issue early alarms. The objective of this review is to present the state of the art in the area of fire detection, prevention and propagation modeling with machine learning algorithms. In order to understand how an artificial intelligence application penetrates an area of fire detection, a quantitative scientometric analysis was first performed. A literature search process was conducted on the SCOPUS database using terms and Boolean expressions related to fire detection techniques and machine learning areas. A number of 2332 documents were returned upon the bibliometric analysis. Fourteen datasets used in the training of deep learning models were examined, discussing critically the quality parameters, such as the dataset volume, class imbalance, and sample diversity. A separate discussion was dedicated to identifying issues that require further research in order to provide further insights, and faster and more accurate models.. The literature survey identified the main issues the current research should address: class imbalance in datasets, misclassification, and datasets currently used in model training. Recent advances in deep learning models such as transfer learning and (vision) transformers were discussed.