Technology audit and production reserves

Development of a hybrid fuzzy decision support system for assessing the effectiveness of artillery fire in conditions of uncertain disturbances

Maksym Maksymov, Oleksiy Kozlov, Oleksiy Maksymov, Ruslan Riaboshapka — Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the processes of determining the effectiveness of artillery fire under conditions of uncertain disturbances, which include wear of the gun barrel, deterioration of the quality of charges and shells of a certain batch. This work addresses the problem of ensuring the adequacy of assessing the effectiveness of artillery fire in cases where the parameters of the gun barrel, the quality of powder charges or shells deviate from the nominal values and are determined inaccurately.

The research used fuzzy logic methods to formalize decision-making processes under conditions of uncertainty, as well as methods of mathematical modeling and statistical analysis to simulate firing sequences and determine effectiveness estimates.

A hybrid fuzzy-logical decision support system (DSS) has been developed and tested, which allows for a comprehensive and highly accurate assessment of the effectiveness of artillery fire. When forming estimates, the DSS takes into account three key parameters that characterize the most significant sources of uncertainty: barrel wear, deterioration of the quality of charges, deterioration of the quality of shells.

The results of computational experiments for various realistic artillery fire scenarios were obtained. In turn, artillery installations with initial barrel wear values of 0.1 and 0.25 were studied when using charges and shells of different quality. During the experiments, it was established that the proposed system provides an adequate, practically useful assessment of the fire efficiency of artillery installations under realistic conditions of uncertainty. In particular, the calculated efficiency values during the entire firing process changed by no more than 12% in the first three experiments and no more than 21% in the next three.

The developed DSS can be used in modern artillery complexes to increase the efficiency of making control decisions, reduce the proportion of misses, save scarce ammunition and reduce the risk of damage to equipment and personnel.

Identification of the characteristics of control signal generation for implementing the traction mode of a physical model of magnetic levitation transport

Thu, 30 Apr 2026 00:00:00 +0300

The object of this research is the process of generating control signals for implementing the traction mode of a physical model of a magnetic levitation transport system.

The problem that was solved was the formation of control signals for implementing the traction mode of the physical model and the method of switching track coils.

In the experimental physical model, the traction mode is implemented by control signals formed on the basis of the angular values of the encoder corresponding to the position of the crew. Based on these signals, the traction modules of the stand are commutated to ensure the necessary polarity of the magnetic field, which creates traction force and ensures the movement of the crew past the section.

The implementation of the traction mode requires accurate determination of the position of the crew relative to the track structure. For this purpose, an encoder signal is used, the information from which is processed in the control unit of the track structure section. A schematic solution and an algorithm for the operation of this unit for a physical model are proposed.

The search for an element base for control boards requires performance verification. From several variations of boards, the one that implements the proposed algorithm with satisfactory process quality was selected.

The main research method is an experiment conducted on a physical model stand. A switching control unit for the traction section of the physical model stand has been developed and implemented, which includes: control boards, a motherboard and software on the Arduino platform. A board option with three DC-DC converters was selected, which ensured stable operation of the traction section.

The main characteristics of electrodynamic processes were obtained, namely: moments of polarity switching of modules, changes in voltage, current and power during the passage of the crew past the section. The switching range of the traction section is approximately five seconds.

The results obtained create the prerequisites for the development of experimental stands and models of maglev transport for further research into traction modes.

Development of a parameter-efficient method for biomedical image synthesis by substituting text conditioning with pathology foundation model embeddings in latent diffusion

Sergii Kuzmin, Oleh Berezsky — Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of synthesizing patches of histopathological images conditioned by embeddings of the pathology foundation model. One of the key problems is that existing approaches to diffusion synthesis either rely on text conditioning via CLIP encoders, which lack morphological understanding, or require full retraining of the generative base model, which requires significant computational resources.

The research used a parameter-efficient adaptation of the previously trained latent diffusion model using low-rank adaptation (LoRA) of the U-Net attention layers in combination with a training MLP projector that reflects the embeddings of the pathology foundation model UNI2-h in the conditioning space of the cross-attention mechanism. Ablation studies of 12 configurations were conducted varying the adapter rank, the number of conditioning tokens, and the projector architecture.

It is confirmed that embeddings of the pathology foundation model can effectively replace text conditioning for the synthesis of histopathology images in a parameter-efficient mode. The optimal configuration achieved FID 77.59 on the validation set and FID 84.17 on the test set when training only 5.53 million parameters, which is 0.64% of the parameters of the base model. This is due to the fact that the proposed method has a number of characteristic features, in particular: embeddings of the pathology foundation model provide morphologically richer conditioning than CLIP-based text representations, and low rank adaptation limits the trainable space to the conditioning pathway.

This provides the possibility of generating histopathology images without text annotations and without full retraining of the model using approximately 12 GB of video memory. Compared to the previous text-conditioned approach on the same dataset, which demonstrated class-wise FID values in the range of 113 to 138, the embedding conditioning method provides significantly higher generation quality while maintaining parameter efficiency.

Identification of influential railway stations using local synchronization in complex networks of train formation plans

Andrii Kyman, Andrii Prokhorchenko, Mykhailo Kravchenko, Mykhailo Muzykin — Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the dynamic processes of coordination of interactions among railway stations within the train formation plan (TFP) network. The problem addressed lies in the insufficiency of traditional topological approaches for identifying influential stations in railway networks. An analysis based solely on degree centrality indicators does not allow the detection of hidden sources of dynamic vulnerability.

An approach is proposed for identifying critical stations and links that reduce the coherence of the TFP network, based on the investigation of local synchronization characteristics. A procedure for analyzing station influence is developed through the integration of centrality measures with a local order parameter calculated using the Kuramoto model. The application of the proposed procedure to real TFP networks representing different structural states revealed changes in local synchronization characteristics. The average value of the local order parameter within the largest strongly connected component decreased from 0.6664 to 0.4976. It was established that topologically significant marshalling stations may exhibit low values of the local order parameter, that is, they may remain locally desynchronized from their immediate neighborhood. It is substantiated that the key factor reducing local synchronization of stations is the phase heterogeneity of their nearest neighborhood, in particular the presence of adjacent stations belonging to other phase clusters.

The practical application of the results is possible provided that the TFP is formalized as a network model and data on the structure of train assignments are available. The proposed approach can be used to support managerial decision-making regarding adjustments to the TFP, improvement of station coordination, and enhancement of the resilience of the railway system under structural changes.

Development of a method for adaptation of radioacoustic sounding systems of the atmosphere to the meteorological conditions

Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of ensuring the Bragg condition between the lengths of acoustic and electromagnetic waves when measuring the altitude profiles of the atmosphere using the radioacoustic sounding (RAS) method.

The problem solved in the work is the lack of a generalized theoretical basis for developing methods for adapting RAS systems to maintain the Bragg condition during the movement of the acoustic wave packet (AWP) in the atmosphere.

In the work, using the theory of stochastic optimal control, a method for frequency adaptation of RAS systems was developed to ensure the Bragg condition along the sounding path. The method includes the operations of estimating the speed of sound, stochastic linear filtering of the AWP state parameter vector and controlling the frequency of the radio signal based on the obtained data. A method for estimating the information parameters of the signal was proposed, and an algorithm for sequential filtering of AWP parameters was developed.

The developed frequency adaptation method will significantly improve the quality indicators of RAS systems – the accuracy of measuring atmospheric temperature profiles and the efficiency of sounding. The use of the method in practice will also allow to increase the range of sounding systems by more effectively adjusting to the Bragg conditions at small values of the signal-to-noise ratio, characteristic of long ranges.

The improvement of the main characteristics of the systems is achieved by more accurately ensuring the Bragg condition in the process of measuring the sound speed values, as a result of which the measurement results do not have systematic errors, and the random component of errors is significantly reduced. Therefore, the averaging time of individual measurement results to achieve the required integral accuracy of estimating the atmospheric temperature profile is significantly reduced, from tens to units of minutes.

The proposed method can be implemented in practice by improving the existing RAS atmospheric systems manufactured by industry.

Development of a decision support model for multi-stage investment decisions in production systems under risk

Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of making investment decisions in production systems under conditions of risk and uncertainty.

In modern enterprise conditions, making investment decisions requires choosing between several options for production development. Their effectiveness depends on possible states of the external environment. A feature of these processes is that investment decisions can have several stages, and their effectiveness depends on the conditions of implementation.

The work focused on developing a decision support model that takes into account the step-by-step implementation of investment projects and evaluates alternatives considering possible environmental scenarios. The analysis showed that traditional approaches are mostly based on one-stage decision models, which limits the ability to consider changes in project implementation conditions.

The model for supporting investment decision-making developed in the research combines single-stage and multi-stage approaches to evaluating the effectiveness of alternatives under conditions of risk. A feature of the obtained results is that they allow determining the expected result of applying multi-stage alternatives and identifying rational investment strategies. An approach to evaluating the efficiency reserve of investment projects in production was also proposed.

During the experimental verification, it was shown that the developed model allows taking into account the staged implementation of alternatives and the information that follows from this. Thanks to this, it provides the possibility of adjusting managerial decisions depending on the actual state of the environment at different stages of the implementation of the adopted decisions.

The developed model can be used in the process of substantiating investment decisions in production systems under conditions of risk.

Development of a hybrid artillery fire control system based on neural networks and uncertainty quantification methods

Yurii Rubel, Yurii Hrytsiuk — Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of controlling the fire of artillery installations in a hybrid ballistic modeling system. The problem addressed lies in the lack of comprehensive studies of systems that combine neural network forecasting with physical iterative refinement and stochastic assessment of projectile dispersion within a single operational pipeline. This paper examines the specific features of developing a hybrid artillery fire control system based on the integration of neural networks, numerical refinement of aiming angles, and methods for quantifying the uncertainty of the ballistic model. A modular system architecture is proposed and investigated, integrating a ballistic simulator with a 4-DOF model in accordance with NATO STANAG 4355. The system is supplemented by a neural network, which generates an initial approximation of the aiming angles. For the subsequent calculation of the aiming angles, an algorithm was implemented using iterative elevation angle refinement via the Brent method and gradient azimuth correction. To assess uncertainty, polynomial chaos expansion (PCE) and Monte Carlo methods were integrated. A synthetic ballistic dataset consisting of 121107 records was generated based on 24 configurations of artillery systems. Validation of the neural network demonstrated a narrowing of the search space for aiming angles to a corridor of ±3–5°, ensuring further rapid convergence of the iterative refinement algorithm. Testing for the 2S22 “Bohdana” artillery system at a range of 20 km showed a deterministic error of 0.68 m. The PCE method achieved an error of 0.47 m, outperforming the Monte Carlo method (5.28 m) by a factor of 11.2. Analysis using the PCE method revealed anisotropic projectile dispersion: σ_x = 168.85 m, σ_z = 80.84 m, CEP50 = 147.3 m. The viability of the hybrid system has been demonstrated under ballistic simulation conditions, laying the groundwork for further validation with real-world firing data.

Development of an approach to forming a frame-based dictionary for the personalization of an educational chatbot

Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of functioning and using AI-based educational chatbots in the educational environment. The problem addressed in the research is the training of chatbots using dictionaries. The paper presents an approach to chatbot personalization through the use of a thematic dictionary and query adjustment with the help of prompts.

A frame-based model of a dictionary has been developed, which can be added to a chatbot as a PDF document. The frames represent the subject domain as a hierarchically organized system. User prompts and the processing of frame structures are integrated with the chatbot through thematic representations. This ensures flexible query formulation, scalability of dictionary resources, and the possibility of further expansion of the subject domain without violating the integrity of the language model.

Schemes for combining contextual projection of query interaction and dictionary search have been developed and substantiated. To implement prompts, an algorithm was designed based on the principle of a “marked bullet” selection of a term or expression. Chatbot personalization is achieved through the formation of a series of user-generated prompts.

Based on the results of experiments on adapting ChatGPT to users’ educational needs, frame-based dictionaries were implemented and tested. For the sequential dictionary implementation scheme, at an accuracy level of 10⁻⁶, the total computational complexity is approximately 4.67, while increasing the accuracy requirement to 10⁻⁸ reduces this value to 3.287. The hierarchical scheme, based on the frame organization of the dictionary and the use of TemaView, demonstrates comparable or lower complexity values (10⁻⁶ = 6.69 and 10⁻⁸ = 4.69).

The practical application lies in supporting the educational process through the use of personalized educational chatbots within learning systems.

Proof-of-Indicators: development and validation of an adaptive consensus mechanism for Internet of Things blockchain networks

Leonid Chepel, Yuriy Boyko — Thu, 30 Apr 2026 00:00:00 +0300

The object of this research is the consensus mechanism in blockchain networks for the Internet of Things (IoT). The expansion of IoT requires a decentralized approach, making blockchain a promising solution. The Proof of Authority (PoA) consensus protocol was identified as the most suitable base for heterogeneous IoT, however, it has limitations regarding fork occurrence, data duplication, and the signer selection process.

The obtained results include the creation of the Proof of Indicators (PoI) consensus protocol that optimizes IoT network by reducing block size and prioritizing capable nodes for consensus tasks for devices with different performance and network conditions. PoI is based on Go-Ethereum's PoA Clique implementation; and a comparative performance analysis was conducted between PoI and Clique in a simulated IoT network.

Testing shows that PoI reduces overall network traffic by 20.5% and decreases network forks by 80%. Under testing, PoI improves transaction throughput and decreases block propagation time, compared to Clique. These gains occur with a modest increase in resource consumption: an average rise of 6.5% in CPU usage and 5.4% in memory usage.

A distinctive feature of this work is combining dynamic node selection with light block propagation within the blockchain consensus layer to address previously found limitations.

The PoI system is a suitable solution for secure and purpose-specific blockchain application in IoT, where blockchain node can be hosted on low-powered devices, such as the Raspberry Pi, creating a fully decentralized cloud-independent infrastructure.

Development of a hybrid method VGG16-FrostNet for adaptive despeckling of synthetic aperture radar (SAR) images using attention mechanism and differentiable Frost filter

Raed Al-Senaikh, Oleksii Rubel — Thu, 30 Apr 2026 00:00:00 +0300

The object of research is the process of suppressing multiplicative speckle noise in synthetic aperture radar (SAR) images, which significantly complicates their analysis. The problem addressed is the lack of end-to-end hybrid methods capable of spatial adaptation by integrating a mathematical model of local statistics (the Frost filter) directly into the neural network computation graph. This research is aimed at automating the process of adaptive SAR image filtering by developing the hybrid VGG16-FrostNet method. These research tasks were addressed by formulating a differentiable mathematical model of the classical Frost filter for integration into a neural network, developing an architecture based on a pretrained VGG16 (Visual Geometry Group) backbone (blocks 1–2), and integrating the Convolutional Block Attention Module (CBAM), which predicts a spatially varying damping coefficient map A_map within 0.5–10.0 for each pixel. The developed hybrid architecture includes a residual branch for detail recovery and was optimized end-to-end using a comprehensive loss function combining L1, Edge Loss (Sobel), SSIM, and attention regularization. The model was trained on synthetic data with gamma-distributed speckle (equivalent looks between 3.0 and 6.0) under typical SAR conditions. On the test set, experimental evaluation yielded a mean PSNR of 34.18 dB and SSIM of 0.97. The gain relative to the noisy image constituted 9.45 dB, and 3.36 dB in PSNR compared to the classical Frost filter with an optimal static coefficient. Edge indicators EPI = 0.8903 and FOM = 0.8340 substantiate reliable preservation of structural boundaries. It was established that the developed hybrid method provides spatially adaptive damping with interpretable attention maps, enabling its deployment in automated SAR data processing pipelines.

Developing a task allocation model for remote health monitoring in smart cities, considering latency, energy consumption, and privacy on fog nodes

Ivan Shevtsov, Tetiana Fesenko — Thu, 30 Apr 2026 00:00:00 +0300

The object of the research is the processes of dynamic distribution of computing tasks in multi-level infrastructures of a smart city. The possibilities of integrating edge, fog, and cloud computing resources for the development of remote patient monitoring systems (Remote Patient Monitoring, RPM) were investigated. The study addresses the challenge of balancing the rapid processing of critical medical signals with the limited energy resources of mobile devices. In addition, the need to ensure the confidentiality of personal data when transferring tasks to third-party fog nodes was addressed through encryption, remote attestation mechanisms, and isolated execution environments.

A comprehensive system model was developed to describe the processes of performing RPM tasks (ECG classification, audio analysis). An offloading strategy was developed, based on a weighted linear to minimize energy consumption and delay. An architectural framework is proposed to ensure the confidentiality of data processing on uncontrolled fog nodes, through the use of Trusted Execution Environment (TEE) technologies and the deployment of Trusted Applications (TA). To validate the solutions, a series of simulations was conducted in the YAFS (Yet Another Fog Simulator) environment to compare Mobile, Hybrid, and Fog scenarios.

It was experimentally established that transitioning to a Fog-oriented strategy results in a radical reduction in the average system latency (from 0.57 s to 0.027–0.030 s). The load on the smartphone is reduced by more than 10 times (from 222–225 mWh to 20.3–20.4 mWh), and the autonomy of wearable sensors increases almost fivefold. It is proven that the use of fog computing provides stable Quality of Service (QoS) on equipment with lower power (500 MIPS). The integration of attestation procedures according to RATS (Remote ATtestation procedureS) standard is intended to enable verification of the integrity of the computing stack before the transfer of confidential data.