Anthropogenic disasters pose a challenge to management in the modern world. At the same time, it is important to have accurate and timely information to assess the level of danger and take appropriate measures to eliminate disasters. Therefore, the purpose of the paper is to develop an effective method for assessing the level of anthropogenic disasters based on information from witnesses to the event. For this purpose, a conceptual model for assessing the consequences of anthropogenic disasters is proposed, the main components of which are the following ones: the analysis of collected data, modeling and assessment of their consequences. The main characteristics of the intelligent method for classifying the level of anthropogenic disasters are considered, in particular, exploratory data analysis using the EDA method, classification based on textual data using SMOTE, and data classification by the ensemble method of machine learning using boosting. The experimental results confirmed that for textual data, the best classification is at level V and level I with an error of 0.97 and 0.94, respectively, and the average error estimate is 0.68. For quantitative data, the classification accuracy of Potential Accident Level relative to Industry Sector is 77%, and the f1-score is 0.88, which indicates a fairly high accuracy of the model. The architecture of a mobile application for classifying the level of anthropogenic disasters has been developed, which reduces the time required to assess consequences of danger in the region. In addition, the proposed approach ensures interaction with dynamic and uncertain environments, which makes it an effective tool for classifying.

The Subject matter of the article is the design and experimental evaluation of an intelligent edge-cloud cyber-physical system for automated inspection of photovoltaic (PV) modules in utility-scale solar power plants based on multi-palette thermal infrared and RGB imagery acquired by unmanned aerial vehicles (UAVs). The Goal of this study is to enhance PV defect detection accuracy and operational efficiency by designing a novel method that converts raw, overlapping sensor data into a compact, geo-referenced inventory of actionable defects. The goal is achieved by systematically increasing detection mean Average Precision (mAP) and reducing false positive rates caused by data redundancy, while simultaneously minimizing bandwidth usage. The Tasks to be solved include: developing a palette-invariant thermal representation that suppresses dependence on pseudo-color rendering and camera internal parameters; fusing this robust thermal stream with contrast-enhanced RGB data using an adaptive gated mechanism that can down-weight unreliable modalities; implementing an on-board active perception loop that re-orients the UAV gimbal for additional views of ambiguous, small-area anomalies; designing a geo-spatial clustering and de-duplication module that merges repeated detections into unique fault events suitable for integration with SCADA and GIS tools; and quantifying the benefits of the proposed architecture on public benchmarks and real-world field trials in comparison with single-modality baselines. The Methods employed include deep convolutional neural networks based on a YOLOv11m-seg instance segmentation backbone trained with a palette-consistency regularization term, gated feature-level fusion of thermal and RGB embeddings, Rodrigues-based calculation of corrective gimbal rotations for adaptive re-acquisition, density-based spatial clustering with the haversine distance metric for geographic de-duplication of detections, and statistical performance analysis using mAP, macro-averaged F1-score, recall, and a duplicate-induced false positive indicator. The following Results were obtained: on the PVF-10 benchmark the proposed system achieves mAP@0.5 = 0.903, exceeding thermal-only and RGB-only detectors by 12–16 percentage points; on the STHS-277 dataset it reaches mAP@0.5 = 0.887; palette-invariant training and adaptive re-acquisition together increase small-target recall to 0.86; geo-spatial clustering reduces the duplicate-induced false positive rate by 12–15 percentage points; field validation at rooftop and ground-mounted plants confirms 96% recall with a low root-mean-square deviation between automatic and manual defect counts; and relevance-only telemetry reduces airborne data transmission by 60–67% while preserving diagnostic fidelity. In Conclusion, the scientific novelty of the results obtained lies in a unified palette-invariant, multi-modal edge-cloud cyber-physical architecture that combines UAV sensing, active perception, geo-spatial reasoning, and bandwidth-aware reporting into a single operational method for photovoltaic module inspection, providing a scalable foundation for condition-based maintenance of large solar power plants.

Current study performs a comparison of the contour inflection point selection methods according to the following criteria: probability of the correct selection, probability of the incorrect selection and an error in the coordinate estimation based on the results of model tasks. Results of the interpolation and differential methods and the method based on wavelet analysis of the contour inflection point were analyzed and graphically illustrated. It is shown that for simple geometric shapes the interpolation method has low precision (it shifts the contour inflection points) while differential method has the best precision properties, however it selects redundant inflection points (it has a low noise resistance) and the method based on the wavelet analysis of the curvature function shows the best results under noisy conditions.

Finding the best products for sale is one of the most important steps in the process of a profitable company creating. That is why the choice of goods for the online store must be made carefully, taking into account both the opportunities and analysis of prospects in the niche, and a number of other important parameters. One of the methods of competitive product choosing can be products analysis in marketplaces based on the emotional feedbacks coloring. Research on product feedbacks is an extremely popular topic, as confirmed by research analysis. Feedbacks can be constantly re-read, but when there are many products in one segment, because there are more and more manufacturers, it is time consuming. Therefore, the development of technology that can automate this process is necessary for the sales business. Paper develops the intelligent method of a competitive product choosing based on the emotional feedbacks coloring, which is divided into three blocks: parser of feedbacks, emotional coloring determination and feedbacks classification. The data will help retailers manage their websites wisely and help customers make purchasing decisions. The implementation of the method was carried out on the data of the Ukrainian site Rozetka, where 4477 feedbacks were used. The classification was tested by eight classical machine- based classification methods, namely Support Vector Classifier, Stochastic Gradient Decent Classifier, Random Forest Classifier, Decision Tree Classifier, Gaussian Naive Bayes, K- Neighbors Classifier, Ada Boost Classifier, Logistic Regression.