Abstract. This research presents an intelligent system for automated generation of domain-specific learning assignments in civil aviation education using fine-tuned T5-small transformer models. Traditional assignment creation requires significant instructor time and expertise in aviation regulations, technical specifications, and safety procedures. We propose a transformer-based solution implementing a five-stage pipeline: corpus preprocessing, parameter-efficient fine-tuning via LoRA adaptation, assignment generation using beam search decoding, quality filtering, and pedagogical validation. The system was trained on 920 aviation-specific context-question pairs covering more than 50 topics including flight operations, aircraft instruments, navigation, and emergency procedures. Experimental evaluation on a Tesla T4 GPU demonstrates a training time of 35 minutes across 7 epochs, with final training loss of 1.3506 and validation loss of 1.221. Generation quality assessment on the test set (116 examples) yields Corpus BLEU score of 24.27, ROUGE-L F1 score of 0.5087, and BERTScore F1 of 0.6017. Aviation terminology coverage analysis shows that 38.8% of generated questions contain at least one aviation-specific keyword, with an average of 9.4% unique aviation terms per question. Additional metrics include a unique bigram ratio of 0.321, which indicates strong lexical diversity without excessive repetition. Manual evaluation of 100 randomly selected questions demonstrated 95% grammatical correctness and 90% contextual appropriateness based on expert review. Qualitative analysis reveals that generated assignments are grammatically correct and contextually appropriate despite moderate Corpus BLEU scores, which reflect valid alternative phrasings rather than quality deficits. Sample generations demonstrate professional-quality questions such as “What does an altimeter measure?” and “What happens when a stall occurs?” The system reduces instructor workload in assignment creation while maintaining technical accuracy and domain relevance, providing a foundation for AI-assisted educational content generation in specialized technical domains.
Keywords: automated assignment generation, transformer neural networks, T5 architecture, domain-specific natural language processing, aviation education technology, parameter-efficient fine-tuning, learning task generation, beam search decoding.
Abstract. In the current global context, supply chains are facing heightened exposure to risks arising from political tensions, restrictive trade measures, and rapid technological shifts. Under these conditions, effective risk management becomes not just a supporting function, but a key element of competitiveness. The present study explores how digital solutions can be applied to identify and reduce the most pressing risks in supply chain operations. The research addresses both operational challenges within logistics processes and broader strategic threats. Its primary aim is to demonstrate the contribution of digitalization to risk reduction and overall resilience. For this purpose, a set of complementary methods was applied: Failure Mode and Effects Analysis (FMEA) to detect weaknesses at the process level and comparative assessment of key performance indicators (KPIs) drawn from real-world cases of information system adoption. The analysis showed that the most significant risks are concentrated in warehouse receiving operations, where manual activities often lead to errors, and in order tracking, where insufficient visibility creates delays and customer dissatisfaction. These issues were identified as top priorities for corrective action. At the same time, the adoption of Warehouse Management Systems (WMS) and monitoring platforms significantly reduced errors, accelerated operations, and lowered costs. Comparative case analysis showed measurable improvements: inventory accuracy increased up to 99%, order picking productivity doubled, and logistics costs were substantially reduced. The study concludes that digital technologies not only address internal vulnerabilities but also create conditions for long-term supply chain resilience. However, external risks such as cyber threats and regulatory changes remain beyond the scope of technology alone and require complementary governance measures. It is therefore recommended to combine technological solutions with organizational practices, enabling companies to both prevent disruptions and build sustainable development strategies.
Keywords: supply chains, risks, digital technologies, warehouse management, efficiency, resilience, information systems.
Abstract. This paper explores modern techniques to process and enhance the reliability of radar signals amid strong noise and interference typical for the secondary surveillance radar (SSR) band at 1030/1090 MHz. Air-traffic radar systems play a critical role in ensuring safe and efficient airspace management, however, their performance degrades under strong noise and interference typical of the 1030/1090 MHz band. A comprehensive analysis of filtering and adaptive signal processing algorithms was conducted to improve the signal-to-noise ratio, stabilize target response parameters, and lower false detection rates. For objective comparison, both traditional and advanced digital filtering methods were evaluated. Special focus was placed on comparing the Median, Butterworth, and recursive Kalman filters, implemented and tested in MATLAB with synthetic data and simulated radar scenarios. The effects of filter order, bandwidth, adaptation coefficients, and sampling interval on signal reconstruction quality were investigated. Using statistical, correlation, and probabilistic analyses, an enhanced adaptive threshold detection method was developed, which accounts for the non-stationary background noise, temporal signal correlations, amplitude fluctuations, and dynamic interference parameters in real time. Results demonstrate that combining and recursively applying these filters greatly enhances the robustness of secondary radars against random and systematic interference, reduces estimation error variance, and improves radar reliability. The practical significance lies in the potential integration of these methods into intelligent air-traffic management systems, which can enhance interference immunity, aircraft identification accuracy, and overall radar surveillance quality in civil aviation.
Keywords: secondary radar, adaptive filtering, Kalman filter, Butterworth filter, median processing, digital signal processing, signal interference, reliability of radar surveillance.
Abstract. The aim of the study is to study the impact of data quality, the factors associated with it, and to formalize the logical and mathematical connections between the arguments of the neural network training function or artificial intelligence to demonstrate the relationship with flight safety.
To achieve the goal, it is necessary to solve the following tasks. The first is to assess the current state of the issue of the use of artificial intelligence in the procedures of pre-flight inspection, as well as repair and maintenance of the aircraft. This is necessary to determine the level of technological integration of neural networks and artificial intelligence. Secondly, based on the data obtained, it is necessary to determine relevant platforms for training artificial intelligence. This will allow you to identify additional technical arguments that affect the final result of AI training. The third is to formalize the logical and mathematical connection between the influencing factors and the final result. Identify additional influencing factors and also formalize. The formal recording method allows for the construction of a procedurally consistent communication line to monitor safety risks.
The following methods were used to solve the problems. An observation method that has been applied to the information that has been collected in the course of tracking the history of the application of various automated optical fault detection technologies. Decomposition, which made it possible to separate the functional part of the computer program that detects malfunctions from the complex technology. Comparative analysis, which made it possible to determine the strengths and weaknesses of various neural networks and the architectures of technical systems for these neural networks and artificial intelligence Mathematical analysis, which allows you to formalize the expressions that characterize the influence of the arguments of a complex function and determine the additivity and multiplication of a complex function. Sabotage and functional analyses, which make it possible to determine the relationship between the arguments of complex functions and the final complex safety function.
As a result, expressions were presented that reflect the logical and mathematical connection in a functionally sequential transfer line from the arguments of a neural network and artificial intelligence to a complex indicator of flight safety.
Keywords: artificial intelligence, neural networks, pre-flight inspection, artificial intelligence models, comparative analysis, flight safety.
Annotation. This article explores the perspective opportunities for securely broadcasting the pulse signals emitted by the ADS-B (Automatic Dependent Surveillance–Broadcast) system, which is currently used in modern aviation systems, against external cyber threats. The objective is to enhance the security level of the existing infrastructure, optimize frequency assessment, and improve traffic management through predictive modeling, thereby enabling more efficient and proactive control. The proposed integration architecture employs deep learning algorithms to analyze aircraft signals and provides functionalities such as signal congestion management, real-time risk forecasting, and proactive prediction of weather and traffic changes. Furthermore, the article presents a performance evaluation of the system operating at 1090 MHz and 978 MHz frequencies and proposes methods for frequency optimization. Research results indicate that incorporating the recognition of device identification via radiometric fingerprints into the ADS-B platform not only enhances security and operational efficiency but also significantly improves the system’s adaptability and responsiveness. This approach opens new avenues for the development of smarter and more predictable future aviation networks.
Keywords: ADS-B technology, aviation safety, frequency optimization, real-time data processing, air traffic control.
Abstract. This article considers the issues of optimization of industrial railway station track distribution in order to improve the efficiency of wagon flow processing. The relevance of the topic is due to the need for rational use of limited track resources, increasing the throughput and processing capacity of stations, and reducing operating costs. The main attention is paid to the development of optimization models and methods that take into account the features of the formation and processing of wagon flows by destination, the schedule, the processing sequence and the limited track fund. The proposed solutions are based on the application of mathematical modeling methods, graph theory, linear and discrete programming. The results can be used to develop automated station control systems, which will significantly improve planning and operational distribution of wagons along the tracks, reduce rolling stock downtime and speed up cargo processing. Objective. To improve the method of distributing industrial railway stations tracks for the accumulation of wagon groups between individual destinations. The optimization problem is to find such a distribution of classification works between an industrial marshalling yard and freight stations, as well as such a distribution of marshalling yard tracks between individual destinations, which ensures minimal time costs for shunting operations.
Keywords: railway transport, industrial railway station, siding, shunting operations, accumulation of wagons, carriage routes.
Abstract. The article presents the general concept of developing a system for automatically lowering the rotor of a wind turbine to protect against strong wind loads. Modern research in the field of automation and optimization of wind turbines shows significant progress. The article provides an overview of the work of a number of scientists on the importance and implementation of adaptive and predictive control methods to improve turbine efficiency and stability. These achievements emphasize the importance of introducing advanced technologies to improve the efficiency and reliability of wind turbines, despite the fact that significant theoretical and practical efforts are required to implement them. Various components were also investigated and analyzed for the development of a linear actuator control system for unloading wind turbines under heavy wind loads. The main focus is on describing the key components of the system, such as an anemometer, a PID controller, a frequency converter, and a linear actuator, as well as their interaction within a mathematical model. The scheme of an automatic control system using a PID controller to lower the rotor of a wind turbine in high wind conditions is analyzed. To bring the control system to optimal parameters, the critical value of the wind speed is calculated using the energy method. The calculation of the critical wind speed is given, at which the protection system is activated. Combining all components of the control system into a single closed-loop control system ensures stable and reliable operation of the wind turbine, reduces the risk of damage and optimizes performance. The proposed concept serves as the basis for further research and development of full-fledged solutions aimed at improving the reliability and efficiency of wind turbines.
Key words: PID controller, anemometer, frequency converter, linear drive, wind generator.
Absctract.This article presents a comparative analysis of uncertainty evaluation methods, GUM (Guide to the Expression of Uncertainty in Measurement) and Monte Carlo, using the calibration of a bimetallic thermometer TBP 63-1 with a measurement range of 0 to 120 °C and an accuracy class of 2.5. The study focuses on the identification and quantitative analysis of key uncertainty factors, including both stochastic and systematic error components. GUM modeling requires a high level of mathematical expertise to perform many procedures, while the Monte Carlo method serves as an alternative for various laboratory studies. A mathematical model is introduced to describe the measured temperature, represented as , where each component reflects its respective contribution to overall uncertainty: is the thermometer reading, is the calibrator correction, and is half the scale division. The analysis process includes calculating the type A standard uncertainty from a series of measurements to assess random fluctuations, while type B standard uncertainties are estimated for systematic sources, assuming uniform and rectangular error distributions. The combined standard uncertainty is integrated to obtain the expanded uncertainty with a coverage factor k=2 and a confidence level of 0.95. Using the Monte Carlo method, 1,000,000 simulations were generated to ensure statistical significance. The analysis results are presented as the final corrected temperature value, accounting for all calculated uncertainties. The primary goal of the study is to determine which method is more effective for this type of measuring equipment and to demonstrate the simplicity of using Microsoft Excel (or similar spreadsheet software) for evaluating measurement uncertainties based on functional dependence.
Keywords: Monte-Carlo method, GUM, bimetallic thermometer, uncertainty, error, Microsoft Excel.
Abstract. The rapid development of digital communication has led to an increase in the number of offensive postings on the Internet. Automatic detection of such content is one of the most pressing problems of our time. However, traditional approaches based on collecting data on a central server can compromise the privacy of personal information. One way to address this issue is to use federated learning. This method involves individual model training on each user’s device without sending data to a central server. In the course of the study, a literature review of scientific papers was conducted and experiences with the federated learning method were analyzed. A special corpus consisting of 73,572 recordings of aggressive and non-aggressive texts was used as a dataset. The DistilBERT model was used to train the model, and the dataset was divided among three clients, each of which trained only their own recordings separately. At the end of each round, the server uses the FedAvg algorithm to combine the model parameters provided by all of the clients on the server to create a common global model. Based on the results, it can be concluded that the federated learning method has two important advantages: first, it works with high accuracy, and second, it ensures the reliability and confidentiality of information.
Keywords: federated learning, natural language processing, DistilBERT, FedAvg, privacy, aggressive content, classification.
Abstract. Trends in the IT sector are constantly evolving, and the ability to anticipate new developments is a key factor in an organization’s success in managing digital resources, including big data. The aim of this study is to explore and analyze big data in order to ensure security in real time. As a result of the bibliometric analysis, 323 scientific articles were obtained from 137 information sources containing the keywords relevant to the study. The research included journal articles, book chapters, and patents. Scientific publications were retrieved from the Scopus and WoS databases using primary and secondary keywords. The study employed the following methods: (a) a systematic literature review based on publications from 2013 to 2023, and (b) a bibliometric analysis of articles published from 2000 to 2023 using the RStudio and Bibliometrix software. Based on the analysis, the authors conclude that in the coming decades, the business landscape of the digital society will be shaped by a strategy focused on processing big data to ensure cybersecurity, grounded in a deep understanding of human behavior and artificial intelligence. This analysis can be used by technical professionals to justify new technological solutions for implementing IT and big data in a digital society, as well as to substantiate recommendations for improving cybersecurity during large-scale processing of big data for applied purposes.
Keywords: big data, digital resources, databases, bibliometric analysis, information technology, data mining.
Abstract. This paper examines the modeling of human evacuation from buildings during fire incidents using macroscopic, microscopic, and hybrid simulation models. The primary objective is to develop strategies aimed at minimizing evacuation time and enhancing safety through the analysis of crowd movement dynamics. The study investigates global flow characteristics using a macroscopic model, individual agent behavior with a microscopic model, and a combined approach in a hybrid model. Technologies such as Unreal Engine (UE) are utilized to create interactive simulations. The scientific significance of this work lies in the comparative evaluation of different modeling approaches for predicting evacuation times. The practical value includes optimizing architectural design, developing effective evacuation plans, and creating VR-based training simulations. Results indicate that while the macroscopic model provides rapid estimation and the microscopic model accounts for individual behavioral factors, the hybrid model most accurately reflects real-world evacuation dynamics.
Key words: Evacuation of people, fire, macroscopic model, microscopic model, hybrid model, simulation modeling, virtual reality, critical situation.
Abstract. The article analyzes the world experience in the development and implementation of robotic mine clearance systems aimed at improving the safety and effectiveness of clearing territories of mines and improvised explosive devices. Modern technologies used in autonomous and remotely controlled systems, including sensor complexes, navigation systems, artificial intelligence elements and integration with unmanned aerial vehicles, are considered. The research methodology is based on a comparative analysis of existing solutions, studying the practice of their application in leading countries such as the United States, Russia, China and the European Union, as well as evaluating the effectiveness of various types of robotic systems, including ground platforms, mobile robots and unmanned aerial vehicles. As part of the practical part of the work, a mathematical model of a combined sensor detection system combining inductive and ultrasonic sensors has been implemented to increase the reliability of object classification during mine clearance. The development is presented in the MATLAB Simulink environment and can serve as a basis for further research and implementation of intelligent sensor modules in domestic robotic complexes. The article also discusses the prospects for localization and adaptation of advanced solutions in Kazakhstan, including the development of domestic robotic platforms, the possibility of using foreign technologies considering the climatic and geographical features of the region, as well as the need to improve the regulatory framework, develop infrastructure and train specialists for the effective implementation of these systems. The importance of further research in the field of autonomous technologies, sensor systems and their integration into the military and humanitarian spheres is emphasized.
Keywords: mine clearance, mines, robotic complexes, improvised explosive devices, autonomous systems.