Ivan Kozhedub Kharkiv National Air Force University

Humanitarian demining requires rapid, accurate and resource-efficient identification of concealed explosive threats. This article presents a multimodal sensor-fusion framework that unites outputs from a visual detector (YOLOv8), two vision language models (ChatGPT--4.1 Vision and Gemini 2.5 Flash Vision) and a custom metal detector into a single integral probability Pfinal. Each modality's raw confidence is calibrated to a comparable probability via temperature or Platt scaling, then fused through learnable non-negative weights or logical rules (OR, AND, k-of-n) to support tunable trade-offs between probability of detection and false-alarm rate. A staged architecture executes preliminary detection on a UAV, confirms candidates at a ground station and escalates cases to the cloud, preserving real-time performance despite limited bandwidth and compute resources.

Active development of optoelectronic systems for unmanned aerial systems has led to both a significant increase in aerial reconnaissance data sets and requirements for the processing of these data sets. Main requirements are the following: ensuring the necessary level of efficiency in processing the ever-growing volumes of aerial reconnaissance data; the need to classify digital images depending on the interests of the aerial reconnaissance system (air reconnaissance area; aerial reconnaissance objects localized on digital aerial photographs); the need to automate individual processes of processing intelligence information in order to reduce the negative impact of the human factor on the decoding process; the need for personnel processing intelligence information to have the necessary skills in using modern tools for intelligent data analysis. A model for classifying digital images generated by on-board optoelectronic systems of unmanned aerial vehicles is being developed in order to automate individual stages of aerial reconnaissance data processing. The essence of the proposed approach is to utilize visual programming modules (blocks) from the “No-Code” platform type “Orange Data Mining” to create a model for classifying digital image arrays in support of aerial reconnaissance. The use of the proposed model allows: to reduce the requirements for the professional skills of personnel processing aerial reconnaissance data through visual programming tools; to provide the possibility of local use (using the UAV command and control station) for processing reconnaissance information; to create conditions for forming data sets for their further annotation and formation of computer vision models in the interests of the aerial reconnaissance system. Further scientific research will be aimed at the integration of the proposed approach to classifying digital images into the preparation of a dataset for developing a model for the automated detection and tracking of aerial reconnaissance objects.

This paper reports the design of a methodological toolkit for financial provision of enterprise sustainability, taking into account the target needs of the enterprise's capital management, available opportunities, and digital changes in the business environment. This study considers the processes and mechanisms of financial provision sustainability of enterprises in the context of digital transformations of the economy. The task addressed relates to the lack of effective methodological tools for flexible adaptation to changes under financial conditions and strategic needs for ensuring financial sustainability of enterprises, taking into account the requirements of balanced economic development. Such tools are based on the mechanism of capital structure optimization and make it possible to build the financial architecture of the enterprise and effectively respond to digital challenges. Using the linear programming method, an economic and mathematical model for optimizing the capital structure of the enterprise has been constructed. It includes an objective function that is focused on maximizing profit and minimizing the weighted average cost of capital and makes it possible to determine the rational ratio of the enterprise's equity and debt capital. The system of constraints of the objective function takes into account the policy of financing the assets of the enterprise under the conditions of digital transformations, and covers the criteria of financial stability, solvency, and efficiency of capital use of the enterprise. This makes it possible to ensure the consistency of financial decisions with the strategic goals of the enterprise's development, to increase its resistance to the latest challenges of the digital economy. The practical significance of the designed methodological toolkit is in the possibility of its application for substantiating management decisions, developing financial and economic policy.

This paper considers the process of segmenting an optoelectronic image acquired from an unmanned aerial vehicle based on the artificial bee colony algorithm. The principal hypothesis of this study assumes that the use of the artificial bee colony algorithm for segmenting an optoelectronic image acquired from an unmanned aerial vehicle could reduce segmentation errors of the first and second kinds. A method for segmenting an optoelectronic image acquired from an unmanned aerial vehicle based on the artificial bee colony algorithm has been improved, which, unlike known ones, involves the following: – initialization of the population of scout bees; – calculation of the objective function; – determining the best and promising positions; – calculation of the optimal value of the segmentation threshold; – image division into segments; – checking the stopping criterion; – bee migration; – acquisition of a segmented image. Experimental studies have been conducted on the segmentation of an optoelectronic image acquired from an unmanned aerial vehicle using a method based on the artificial bee colony algorithm. The visual quality of the segmented image makes it possible to conclude that segmentation using the artificial bee colony method is possible. Comparative analysis of segmented images (improved and known methods) indicates a clearer separation of the object of interest (car) using the method based on the artificial bee colony algorithm. The results of calculating segmentation errors of the first and second kind indicate a reduction in segmentation errors of the first kind by 9% and errors of the second kind by 7% when segmenting an optoelectronic image using the method based on the artificial bee colony algorithm

Research objective is to estimate the parameters of extreme value distribution laws constrained by parameters using the maximum likelihood method and to model the conditions of their occurrence with STATGRAPHICS and MATLAB tools. The subject of the study includes Anscombe's quartet and the first (Gumbel), second (Fréchet), and third (Weibull) laws of extreme value distribution. The research method involves numerical methods for solving systems of equations obtained by the maximum likelihood method, as well as statistical modeling techniques. The results of research show, through the example of statistical data analysis of Anscombe's quartet, the necessity of verifying the correspondence between the physical content of the studied processes and the applicability of extreme value distribution laws for their analysis. The linear regression equation, which corresponds to all possible combinations of this quartet’s data, is not optimal according to the criterion of maximum coefficient of determination. Using this criterion, it has been established that different data pairs have different nonlinear regression equations. Ignoring this fact may lead to errors in managing the processes they model. It has been shown that the data presented in Anscombe's quartet follow Gumbel's law, although the construction scheme of Anscombe's quartet does not correspond to the conditions of its occurrence. The limiting laws of extreme value distribution are presented in a form convenient for practical application in the design of engineering structures with constraints on the parameters of these distributions, caused by the specifics of the design. Parameter estites of extreme value distribution laws were performed using the maximum likelihood method. A numerical method for solving the corresponding systems of equations was considered. For modeling the scheme of extreme value occurrence, a matrix of random variables was generated, with twelve columns simulating monthly observations of a certain geophysical phenomenon, and one hundred rows representing a century of observations. The possible distributions of the elements of these matrices (initial distributions) were assumed as follows: double exponential distribution, Laplace distribution, lognormal distribution, Rayleigh distribution, normal distribution, Champernowne distribution, maximum extreme value distribution, minimum extreme value distribution, Birnbaum-Saunders distribution, Burr Type XII distribution, generalized extreme value distribution, inverse Gaussian distribution, Weibull distribution. The conditions of invariance of the extreme statistics distribution laws concerning the initial distributions were confirmed in only 12 out of 48 possible cases when modeled with STATGRAPHICS, and only in 5 out of 18 possible cases when modeled with MATLAB. The modeling results revealed a significant difference between the actual and theoretically possible extreme value distribution laws, which may be due to the peculiarities of the random number generation algorithms and the choice of the best-fit distributions used in STATGRAPHICS and MATLAB systems.

Background.. We use the mechanical action of laser radiation in the optical range for levitation, holding and moving of microparticles. We use that fact that the radiation beam presses on the particle, pulls it into a region of high field intensity and holds it there. We use the effect of strong absorption and scattering of microwave radiation by very thin conductors - metal wires, graphite and semiconductor fibers, to overcome difficulties caused by comparable large focal spot of the beam in the microwave range. Objectives. To provide an analysis of the possibility of levitation and control of the movement of thin metal targets, which can be metal conductors with a diameter of several micrometers and a length of several millimeters, using microwave radiation without focusing on the target. Results. We provide a theoretical analysis of the effect of microwave radiation strong pressure on thin conductors. We derive a condition for the maximum of radiation pressure - the relationship between the radiation wavelength, the diameter of wire and its conductivity. Also, we provide results of measurement of forces acting on thin conductors. For such purpose, we have used a torsion bar scales with a wire suspension with a diameter of several micrometers and an optical rotation angle reading. We show the good agreement between the calculation and experimental results.

The object of the study is the process of choosing the optimal strategy for assessing a military serviceman. The problem, which was being solved, is non-adaptability of the existing scientific approach to the necessity of taking into account the rapid transformation of functional responsibilities and long-term career planning when assessing military servicemen. In terms of improving the approach of choosing the optimal strategy for assessing a military serviceman, the algorithm for choosing the appropriate strategy has been suggested. The algorithm is based on the application of optimal selection criteria (Laplace, Wald and Savage). The essence of the developed algorithm implies the level of uncertainty regarding the possible transformation of service functions, the possible transformation of service functions and acquired competencies. The peculiarity of the developed algorithm is that it allows to take into account the long-term perspective of career development by considering the uncertainty in the process of transformation of service functions. As part of improving the approach to choosing the optimal strategy for assessing a serviceman, the procedure for choosing the appropriate optimal strategy has been improved. The peculiarity of the suggested procedure is that it is based on the identification of the level of uncertainty, the use of a tabular method for determining the correspondence of service functions to competencies, and the distribution of competencies on the basis of their acquisition. This procedure allows to increase the accuracy of determining the level of correspondence of a military serviceman’s competencies to service functions by 15% if no changes occur, and by 17% in case of minor changes and by 24% in case of fundamental changes. A distinctive feature of the procedure is the division of competencies into three groups: professional, general development, and acquired through experience. The scope of practical application of the improved procedure is the management processes to ensure the effectiveness of HR management in the command-and-control bodies of military formations

The article focuses on the study of learning strategies aimed at acquiring professional vocabulary since professional lexical competence is an integral part of professional competence, the linguistic basis of foreign language communicative competence, and higher education institutions make great efforts in professional communication. Growing demands on professional language training of specialists in technical specialties, expansion and strengthening of international contacts, and growth of scientific progress determine the need for proficiency in English for Specific Purposes at a sufficiently high level. These trends make it necessary to search for optimal methods for intensifying teaching professionally oriented terminology and effectively acquiring terminological units. Deepening into the theoretical background of the concepts of "professionally oriented lexis" and "learning strategy", the article highlights the available categorization of learning strategies. Also, it emphasizes scientists’ various approaches to interpreting these concepts. The article clarifies the didactic possibilities of learning strategies and determines possible ways of their implementation in teaching/learning a foreign language in the context of higher education. The article provides an understanding of the learning strategies' impact on the dynamics of acquiring professional vocabulary, emphasizes the importance of following the basic principles of professional lexis teaching, focuses on the teacher’s role not only in instructing but also in forming and developing foreign language learning strategies in students. The article details the ways of implementing cognitive learning strategies in English classes. It gives examples of exercises that can significantly improve learning effectiveness, using cognitive strategies and creating favorable conditions for acquiring specialized vocabulary.

The article discusses the problematic issue of the large time spent by a technical specialist when searching for options to provide communication to a unit in a new area. This paper analyzes commercial and non-commercial tools from both global companies such as Google, Ubiquiti, and special developments based on state military institutes. All of the considered tools allow technical personnel to find ways to solve the problems of building radio relay communication lines, but require professional knowledge and experience, which significantly increases the impact of the human factor on the result. A conceptual model has been developed that can be used to create software that will reduce the time spent analyzing the data required for decision-making, reduce human involvement, which in turn will increase the efficiency of a potential communication system. This data includes a topographic map of the area, the existing telecommunications network, a list of own mobile communications equipment, and the power grid. At the same time, the proposals provided by the program at the planning stage can be evaluated by the level of efficiency of the task according to the criterion of the best quality of the communication channel with the least involvement of communication means. User feedback will allow the built-in artificial intelligence to learn and provide better suggestions in the future. The purpose of this paper is to develop a conceptual model for the automatic construction of a communication system based on the existing telecommunication and topographic situation of the area, consider the possibilities of deploying (increasing) own (donated) means (equipment) of communication within the framework of the task. An additional extension of the model may be to take into account information about the power grid available in the area, which will reduce the financial costs of maintaining the communication system in constant operation (readiness).

The object of this study is the process of forming a minimally sufficient set of reference images for use in correlation-extreme navigation systems when changing the navigation parameters of unmanned aerial vehicles. The paper reports the results related to solving the task to form a set of reference images taking into account changes in the navigation parameters of high-speed unmanned aerial vehicles and their impact on the information features of images. The effect of changing the viewing angles and altitude on the formation of segmented images by energy characteristics has been studied. The discrete steps for navigation parameters were established, at which the correlation between image fragments is maintained at the level of 0.9. These values are from 90 to 120 meters in height and from 15° to 25° in angular parameters. The effect of the structure of segmented images on the selection of the reference object has been studied. It is shown that the feature of the selection of the reference object in the segmented image is the value of the fractal dimension 2.998…2.999. The study was conducted in the MATLAB software environment using the source image selected from Google Earth Pro. The application of the selected sequence of constructing fragments of reference images has made it possible to identify objects that have the best characteristics in terms of signal-to-noise ratio and structure with increasing discreteness of navigation parameters. The method differs from known ones in using the image structure as information features along with the brightness and contrast of objects. This would reduce the number of fragments of reference images while maintaining the accuracy indicator. The results could be implemented in secondary processing systems of correlation-extreme navigation systems used on high-speed unmanned aerial vehicles