No. 3 (2026)

This paper examines the problem of improving the detection of computer attacks (CA) by an intrusion detection system (IDS) under conditions of significant network traffic data imbalance. Based on an analysis of methods for reducing data imbalance, it is concluded that classical methods of balancing and generative augmentation do not preserve the statistical structure of multidimensional tabular data, including their fractal properties and self-similarity, which reduces the quality of classifier training. This paper proposes a method for detecting computer attacks (CA) based on the H-DDPM data augmentation model, a modification of the DDPM diffusion probabilistic model, in which the variance of the added Gaussian noise in the forward process depends on the Hurst exponent H for each CA class. The method includes data preprocessing, the formation of time series using sliding windows, H estimation using DFA and R/S methods, and the generation of synthetic data for training the LSTM classifier. The method is evaluated using the general performance metrics Accuracy, Recall, F1, ROC-AUC, and G-means, as well as Precision, Recall, and
F1-score for each class. Experiments were conducted on the CSE-CSE-CIC-IDS2018 and UNSW-NB15 datasets. A comparison was made with other methods, such as SMOTE, GAN, and DDPM. The experimental results show that H-DDPM improves the efficiency of CA detection, outperforming similar methods in terms of imbalance-sensitive metrics. Furthermore, experimental validations demonstrate that directly using the Hurst H exponent for CA classes in the H-DDPM model improves the recall and balanced quality of CA detection. It is noted that H-DDPM has an impact on CA classification, manifested by an increase in false positives and a decrease in the ROC-AUC metric, which requires additional tuning of the classifier model hyperparameters and filtering of synthetic data

This study aims to develop a formalized information security threat model for hyperconverged infrastructure by constructing multi-component attack scenarios using the EPC (Event-driven Process Chain) methodology, taking into account the architectural features of HCI, the cascading nature of threat propagation, specific security objects, and the multitenant access model. The method of system analysis, the principle of the ARIS (Architecture of Integrated Information Systems) methodology, as well as the method of formalization of multicomponent attack scenarios are applied in the work. The technical basis was the FSTEC of Russia's methodology for assessing information security threats, the register of information security threats, and vulnerability information from the FSTEC Threat databank and the International Database (NVD). The study revealed the key architectural features of hyperconverged infrastructure as an object of protection close integration of components (computing, storage, and network), software definability of components, multitenancy, cascading nature of threat propagation. A classification of violators by privilege level is proposed. Two formalized scenarios of multicomponent attacks on specific threat targets in HCI have been developed. The scenarios are presented in the form of EPC diagrams, which allows not only to visualize the actions of the violator, visually present the tactics and techniques used by him, but also to numerically assess the probabilities of the scenarios. Based on the study of the structural and functional characteristics of HCI and taking into account the FSTEC threat assessment model, a generalized threat model of hyperconverged infrastructure with examples of filling for some affected objects has been developed. The scientific novelty of the study lies in the adaptation of the EPC threat modeling method for hyperconverged infrastructure, taking into account its architectural features and multitenancy, as well as in the development of formalized attack scenarios reflecting real vulnerability exploitation chains published in 2025-2026

The integration of large language models (LLM) into geographic information systems (GIS) opens up new opportunities for spatial analysis, but it is accompanied by specific vulnerabilities such as "hint injection" (prompt injection). Such attacks allow attackers to bypass LLM security mechanisms, manipulate issuance, gain access to confidential information, and violate data integrity. Using space allows you to access an object not directly, but through its spatial relationships with other objects. Existing keyword or template filtering methods do not provide reliable protection due to the constant emergence of new attack scenarios. This determines the relevance of developing adaptive, self-learning algorithms for filtering queries for industrial injections to large language models. The aim of the study is to develop an algorithm for filtering prompt injections for LLM, based on the Case-Based Reasoning (CBR) method. The paper proposes an algorithm for comparing LLM queries with a database of previously known promt injections. The experiment showed that as the database of use cases accumulates, the accuracy of detecting prompt injections increases from 42% to 83%. At the same time, the processing time for a single request increases slightly (from 0.18 to 0.19 seconds with a 23% increase in the database). Approaches to generalizing the precedent base and introspection of the precedent base were also proposed. The proposed algorithm makes it possible to increase the security of LLM-interface systems against prompt injections due to adaptivity and self-learning. The practical significance lies in the possibility of implementing the developed filter into information systems to prevent leaks and manipulation of spatial data. Further research is related to the development of methods for automatic generalization of use cases and the integration of additional contextual analyzers.

This paper addresses the fundamental challenge of overcoming the systemic epistemological crisis in digital forensics. This crisis is driven by an expanding semantic gap between the probabilistic and stochastic nature of digital traces—frequently compromised by anti-forensic techniques—and the rigorous demands of adversarial legal proceedings for the legal certainty of evidence. The article is conceptual in nature and establishes the theoretical foundation of computational forensics as an independent scientific discipline. The study justifies a necessary paradigm shift from traditional heuristic artifact-discovery approaches and subjective expert opinions toward a rigorous methodology grounded in the principles of algorithmic reproducibility, the measurability of uncertainty, and formal verifiability. The author develops a set-theoretic ontological model of a computer incident, which is built upon the "subject–method–object" (S-M-O) triad and the axiom of process trace conservation. This model enables a one-to-one mapping of low-level technical indicators (IoA, IoB, IoC) onto the legal elements of a crime (corpus delicti). Furthermore, a methodology for the formal verification of hypothesis validity is proposed, incorporating the criteria of structural completeness, causal coherence, logical consistency, and factual grounding. For the first time, a mathematical model for assessing the reliability of expert findings using a logistic function (trust function) is introduced into scientific discourse. This model enables the calculation of the probability of a juridical fact by aggregating taxonomic compliance metrics, the strength of causal relationships within the incident graph, and an environmental entropy penalty. The application of the developed approach transforms forensic incident reconstruction from an ill-posed inverse problem into a deterministic procedure, ensuring the mathematically provable objectivity of the evidentiary base even under conditions of incomplete data and active anti-forensic countermeasures.

Relevance. The effectiveness of correlation in SIEM is determined not only by the completeness of incoming events, but also by their suitability for scenario-based analysis. Practical logging defects, including under-logging, violations of temporal consistency, loss of mandatory attributes, and record duplication, degrade detection quality and increase the likelihood of false positives. Objective. To develop a scenario- and risk-weighted methodology for the quantitative assessment of logging deficit as a measure of the suitability of an event stream for correlation. Tasks. To form a scenario-justified set of controlled event classes, assign risk weights to them, construct a probabilistic baseline intensity profile, introduce criteria of temporal validity and structural suitability of records, and obtain an integral indicator with diagnostic interpretation of the causes of logging quality degradation. Methods. The study employs scenario analysis of detection content, risk-weighted aggregation, baseline construction from historical data, monitoring of event temporal validity, and assessment of the completeness of minimally sufficient profiles of correlation-significant attributes. Results. The proposed method extends the basic logging deficit indicator by introducing a probabilistic baseline intensity profile, sensitivity to over-logging, loss of correlation-significant fields, and a diagnostic decomposition of logging quality degradation causes. Experimental validation showed that in the baseline mode, DL_new = 0.000 while DL_old = 0.297, confirming the absence of false positives produced by the new indicator on a normal event stream. In the duplication scenario, DL_old = 0.000 while DL_new = 0.073, demonstrating the ability of the new inidcator to detect over-logging that remains invisible to the basic indicator. When temporal consistency is violated, all indicators reach 1.000, which corresponds to the complete unsuitability of the event stream for correlation. Conclusions and significance. The scientific novelty lies in the formalization of logging deficit as a deficit of event-stream suitability for correlation, taking into account the scenario significance of event classes and introducing a diagnostic decomposition that makes it possible to identify the dominant diagnostic component of degradation. The practical significance lies in the ability to quantitatively justify priority measures for improving event sources, normalization procedures, and correlation rules in SIEM.

It is widely accepted that symmetries are rarely, if ever, exact in any biological system. Minimum imperfections and external perturbations are inevitable, meaning some degree of symmetry breaking must always be assumed present. However, within the field of computational neuroscience, only a limited number of studies have focused on the intrinsic dynamics of neural network models under conditions of imperfect symmetry. This paper directly addresses this gap by exploring the profound effects of deliberate symmetry breaking on the dynamics of the Tabu learning neuron (TLN) model. By introducing and varying a small symmetry perturbation parameter, we precisely control the model's symmetry and disclose a rich repertoire of previously unreported asymmetric dynamic behaviors. These include complex regimes of asymmetric multistability, where multiple distinct stable states coexist, and asymmetric chaotic bursting oscillations. Furthermore, we reveal the presence of several metastable or transient phenomena, such as transient asymmetric bursting oscillations and transient chaos, where the system exhibits complex behavior before eventually settling into a simpler attractor. These intricate dynamic features are systematically illustrated and verified by utilizing a suite of nonlinear investigation tools. Our methodology encompasses the analysis of time series, bifurcation diagrams, and one- and two-dimensional plots of the leading Lyapunov exponent, complemented by detailed plots of state space trajectories. The plethora of novel behaviors uncovered in this study, stemming directly from controlled symmetry imperfection, represents a significant contribution to the fields of nonlinear dynamics and neuromorphic modelling, highlighting the critical role of asymmetry in generating computational complexity.

This paper investigates the application of classical machine learning algorithms for signal demodulation in a high-frequency communication channel described by the Watterson model. The relevance of this task stems from the need to improve noise immunity during broadband data transmission under conditions of multipath propagation and Doppler distortions. The classifiers used include Random Forest, Decision Tree, and k-nearest neighbors (KNN). The models are trained on synthesized data generated in the MATLAB Simulink environment, with varying path delays (0–4 ms) and Doppler shifts (0–10 Hz).
The input features comprise the coordinates of signal constellations as well as the history of received symbols, since intersymbol interference significantly affects the position of the current point. The experiments revealed a critical feature: demodulation algorithms exhibit high sensitivity to variations in Doppler shift. A deviation of just 0.1 Hz from the training set parameters alters the structure of the signal constellation, increasing the bit error rate (BER) to 0.5. When the Doppler parameters align with the training grid, the Random Forest algorithm demonstrates the best performance, achieving a BER <0.01 on the full dataset. To address this sensitivity, a local training method with a reduced Doppler shift sampling step (down to 0.1 Hz) is proposed, enabling a BER ≤0.01 for Random Forest. Due to the exponential growth of the training dataset size (up to 1000 GB), data reduction techniques were developed: excluding time parameters and sampling data based on possible combinations of preceding bits. This reduced the data volume by a factor of 31.5 while maintaining a BER <10% for Random Forest with a training step of 0.2 Hz. The study concludes that the Random Forest classifier is suitable for demodulation in HF channels and highlights the necessity of adapting the training step to the required Doppler shift accuracy.

Intruder recognition in uncontrolled environments is a critical function of biometric-based security control systems (SCS), ensuring protection at facilities with large crowds. The accuracy of such systems can be improved by combining multiple biometric traits extracted from video imagery. However, processing video data involves several challenges, including viewpoint variation, occlusion, and selecting an appropriate feature fusion level. To resolve these challenges, a complete approach to intruder recognition is proposed. The approach is based on video preprocessing, the combination of two convolutional neural networks (CNNs), Dempster–Shafer theory, and the random forest method. These techniques fuse behavioral biometric features at the score level to classify video sequences. Gait and gestural behavior are selected as biometric modalities, as they can be captured without direct subject interaction. For the classification task, three classes of video data are defined: class 1 – person is not intruder, class 2 – a potential intruder, class 3 – intruder. The study also presents a general architecture for the proposed approach, along with a detailed description of its processing stages. The effectiveness of the approach is measured through experiments performed on the KTH dataset, which comprises six types of simple human actions performed by different subjects under different background conditions. Experimental results show that the proposed approach improves intruder recognition accuracy in uncontrolled environments, achieving an 87 % classification rate.

The article addresses the problem of reliable face recognition in critical areas such as video surveillance and biometric authentication under low-light conditions. Existing approaches typically separate the tasks of image enhancement and face identification, which leads to error accumulation and loss of informative features. The aim of this work is to overcome this limitation by developing and theoretically substantiating a hybrid end-to-end architecture in which image enhancement and face recognition are solved jointly. The study provides a systematic review of modern methods, including classical algorithms (such as histogram equalization and noise suppression) and advanced deep neural networks (including EnlightenGAN, Zero-DCE, ArcFace, and RetinaFace). The main contribution is the integration of generative and identification modules into a single computational graph. The key result of the study is the demonstration that joint optimization of all processing stages within a unified model, unlike fragmented solutions, fundamentally changes the approach to the problem. Theoretical analysis and comparative evaluation of existing concepts show that the proposed architecture ensures a more efficient gradient flow during training, leading to the formation of higher-quality and noise-robust identity features. It is shown that this approach prevents error accumulation between stages and minimizes information loss. The novelty of the work lies in the holistic, end-to-end view of the face recognition problem under low-light conditions. The practical significance is confirmed by the applicability of the architecture in real systems, where its implementation can potentially improve reliability and processing speed by combining heterogeneous tasks into a single optimizable framework.

The article addresses the problem of reliable face recognition in critical areas such as video surveillance and biometric authentication under low-light conditions. Existing approaches typically separate the tasks of image enhancement and face identification, which leads to error accumulation and loss of informative features. The aim of this work is to overcome this limitation by developing and theoretically substantiating a hybrid end-to-end architecture in which image enhancement and face recognition are solved jointly. The study provides a systematic review of modern methods, including classical algorithms (such as histogram equalization and noise suppression) and advanced deep neural networks (including EnlightenGAN, Zero-DCE, ArcFace, and RetinaFace). The main contribution is the integration of generative and identification modules into a single computational graph. The key result of the study is the demonstration that joint optimization of all processing stages within a unified model, unlike fragmented solutions, fundamentally changes the approach to the problem. Theoretical analysis and comparative evaluation of existing concepts show that the proposed architecture ensures a more efficient gradient flow during training, leading to the formation of higher-quality and noise-robust identity features. It is shown that this approach prevents error accumulation between stages and minimizes information loss. The novelty of the work lies in the holistic, end-to-end view of the face recognition problem under low-light conditions. The practical significance is confirmed by the applicability of the architecture in real systems, where its implementation can potentially improve reliability and processing speed by combining heterogeneous tasks into a single optimizable framework.

Managing student mobility amid demographic fluctuations and the digitalization of higher education is becoming a key factor in university sustainability, affecting both financial performance and the quality of the educational process. The increasing complexity of processes such as admissions, withdrawals, academic leaves of absence, transfers, and reinstatements requires a shift from expert assessments to formalized models and predictive analytics based on the processing of large datasets. The aim of this study is to develop and evaluate the effectiveness of a model for forecasting student population dynamics based on machine learning algorithms and using time series analysis. Aggregated statistical data on student mobility at Russian and Chinese universities for the period 2013–2024 were used as the empirical basis, which allowed for consideration of both the structural features of national higher education systems and long-term trends and anomalous events (including the impact of the COVID-19 pandemic). The methodological framework includes a dynamic student cohort balance model in the form of a system of recurrent equations describing transitions between academic years and enrollment statuses, and an additive Prophet model used for independent forecasting of key flows (admissions, transfers, withdrawals, academic leaves of absence, reinstatements) as separate time series. The software implementation is based on the FastAPI–React stack, utilizing the SQLAlchemy ORM layer and mechanisms for caching the results of predictive calculations, which ensures high performance when processing queries. Experimental results on real data demonstrate the robustness of the developed model to nonlinear changes in the input series and confirm the feasibility of integrating machine learning into the student movement management system. The practical significance of this work lies in the creation of an information and analytical system that provides automated monitoring and forecasting of student movement trajectories between courses and statuses, enabling universities to transition from reactive to proactive planning of admissions campaigns, classroom allocation, and the distribution of personnel and infrastructure resources.

This paper focuses on overcoming the systemic limitations of the large-scale language model (LLM) scaling paradigm, which are related to data exhaustion and exponential growth in computational costs. This enables the development of more efficient approaches to building NLP models without sacrificing their performance. The goal of this study is to compare the performance of a standard transformer architecture (nanoGPT) and a model using semantic embeddings (nanoSonar) for language modeling tasks under resource constraints. Working with conceptual embeddings allows us to identify deeper linguistic patterns and reduce the amount of required training data, significantly improving modeling efficiency. The study utilized the TinyStories dataset, which includes short narratives with a clear structure. Before implementing the models, the data was preprocessed: for nanoGPT, tokenization was performed using the BPE method, and for nanoSonar, text was converted into semantic embeddings using a pretrained Sonar model. The models were evaluated using the loss and perplexity metrics. The results showed that the nanoSonar model provides significantly lower perplexity (6.609 versus 39.151 for nanoGPT) and demonstrates more robust training dynamics at later stages. This paper presents an analysis of modern approaches to scaling optimization (MoE, distillation, PEFT) and promising architectures (LRM, SSM, RWKV), and provides practical recommendations for applying models operating in the space of semantic embeddings to domain-specific problems and systems with limited computational resources. The results of this study can be useful in developing efficient language models that combine high generation quality with a cost-effective architecture.

This paper explores a method of neurolinguistic text identification aimed at analyzing and verifying information sources, including texts generated by artificial intelligence systems. Three versions of the information states of the text of Luo Guanzhong's Romance of the Three Kingdoms are analyzed: the original text and the text generated by the Gemini and GPT artificial intelligence systems. The study aims to formulate and substantiate parameters for use as identification factors in the generated text, as well as to create 3D images of neurolinguistic textual identification of information sources. The specialized software package "Neurolinguistic Text Identification Analyzer" is used, processing text data based on horizontal and vertical scanning of neurolinguistic information frames. As a result, information spectra, quantitative characteristics (information capacity, entropy, redundancy), and 3D neurolinguistic information images of neurolinguistic frames of textual information of the Chinese work are formed. A comparison of the identity levels of neurolinguistic 3D informational images of the textual information source and neurolinguistic information frames shows that the highest level of identity is observed when comparing the texts of neurolinguistic information frames with the original text, while the lowest level of identity is observed when comparing the original text with the text generated using neural networks. The obtained results demonstrate significant differences between the parameters of the neurolinguistic information frames of the original text and the parameters of the text generated by neural networks, both in terms of quantitative text characteristics and the characteristics of the neurolinguistic 3D informational images. It was found that the neurolinguistic 3D informational images of texts generated by neural networks have a smoother visual representation structure and an excellent color distribution compared to the neurolinguistic 3D images of the original text. The practical significance of this study lies in the application of an approach that allows for the identification of generated text and the verification of information sources. The obtained results open up prospects for further work and the possibility of creating programs capable of detecting the presence of text generation

The relevance of this study is determined by the growing interest in hyperspectral imaging as a tool for non-destructive testing of building materials and structures, as well as by the insufficient systematization of modern methods for processing such data. The aim of the work is to provide a systematic analysis of HSI processing algorithms, identify their advantages and limitations, and determine the prospects for their application in construction diagnostics. The study examines the specific features of hyperspectral data, including high dimensionality, noise, calibration errors, atmospheric distortions, and the shortage of labeled datasets. The evolution of approaches is shown: from classical machine learning methods and manual feature engineering to deep neural networks. Dimensionality reduction methods, kNN classifiers, Bayesian models, logistic regression, Random Forest, SVM, and MLP are analyzed, along with methods for incorporating spectral-spatial context. Special attention is paid to modern deep learning architectures: 1D, 2D, and 3D CNNs, RNNs, LSTM/GRU models, hybrid CNN–RNN models, transformers, and CNN–Transformer schemes. Transfer learning, semi-supervised learning, self-supervised learning, few-shot learning, meta-learning, and domain adaptation are considered separately as ways to overcome the limited availability of labeled data. The approaches are compared in terms of data requirements, computational complexity, robustness to noise, and their ability to account for spectral and spatial dependencies. It is shown that the most promising models for construction diagnostics are hybrid models that combine local convolutional features, the global context of attention mechanisms, and the possibility of fine-tuning on small specialized datasets. The paper summarizes the current state of the field and forms a basis for selecting methods for defect detection, moisture assessment, corrosion analysis, and evaluation of degradation in building structures. The conclusions formulated in the study can be used when designing experimental protocols and selecting architectures for further applied research in the field of building monitoring

The article presents a hybrid method for predicting the direction of bond price movements, combining the XGBoost machine learning method with hyperparameter optimization using a genetic algorithm. The research is aimed at solving the problem of binary classification of the direction of the price of the Russian Railways bond on the next trading day. The research methodology includes the formation of an expanded feature space of 18 technical indicators calculated on the basis of daily OHLCV data. To configure XGBoost hyperparameters, a genetic algorithm is implemented using the DEAP library. The study was conducted on three time horizons: 01.01.21-31.10.25, 01.01.23-31.10.25, from 01.01.24-31.10.25.
As a result, a significant dependence of the effectiveness of the model on the time horizon of the training data is shown. The best quality was demonstrated by a model trained on data from 2024-2025, with an accuracy of 64.4% in the test sample, balanced precision and recall metrics, as well as high F1-score scores for both classes. Models trained over longer periods (2021-2025 and 2023-2025) showed a decrease in generalizing ability, which indicates that the relevance of the data prevails over its volume in the context of changes in Russia's monetary policy in 2021-2025. To maintain the predictive power of the model in changing market conditions, it is recommended to use a sliding learning window of 1.5–2 years. The comparison with the "Buy & Hold" strategy confirmed the effectiveness of the proposed hybrid approach

The development of intelligent warehouse systems and the automation of logistics processes require effective solutions for task allocation in heterogeneous multi-robot complexes, particularly in the cooperative transportation of large and heavy cargo. The aim of this work is to develop and verify a hybrid model for cooperative transportation task (CTT) identification and allocation in a warehouse environment, taking into account multi-criteria optimization. A brief review of publications on the application of mivar technologies and machine learning methods in the mathematical modeling of complex robotic systems is provided. A two-level approach is proposed, including a mivar decision-making system for the automatic identification of CTT and a task allocation model based on a combined auction algorithm. The required number of transport robots (RT) is determined by the mivar decision-making system, considering the size and mass of the cargo. The developed mathematical model for CTT allocation aims to improve efficiency and reliability by accounting for key dynamic factors (heterogeneity, redundancy, and path-dependent costs). Simulation experiments with 30 transport robots and 100 tasks demonstrated the superiority of the proposed method over baseline strategies (Random, Nearest Neighbor, Greedy Capacity): when processing 6 CTT, the total cost reduction reached up to 40.7%, and with 12 tasks, an additional reduction of 10.8% was achieved while maintaining 100% success rate. The model’s ability to scale efficiently was established, manifesting in an additional cost reduction of 10.8% as the number of tasks increased. The results indicate the robustness, adaptability, and high practical applicability of the model for integration into modern intelligent warehouse systems that handle a diversified range of cargo

The paper presents the results of a study of the internal boundary value problem for a mixed hyperbolic-parabolic equation characterized by the presence of multiple characteristics. The domain in which the equation is considered is finite and simply connected, consisting of two parts: hyperbolic and parabolic. This division of the domain is determined by the specific nature of the equation under consideration. The boundary between the hyperbolic and parabolic regions is a degeneracy line, where the transition from a hyperbolic to a parabolic equation occurs. A key feature of the problem is the boundary condition specified in the hyperbolic part of the region. This condition relates the values of the desired function at points belonging to the characteristics of the region, which adds significant complexity to the problem. The purpose of this work is to prove the existence and uniqueness theorem for the given boundary value problem. The proof is conducted for various cases determined by the discriminant of the cubic characteristic equation associated with the partial differential equation. Different values of the discriminant indicate different qualitative properties of the solutions and require an individual approach to the proof.
To prove the uniqueness of a solution, the energy integral method is used. This method allows one to establish certain energy inequalities that relate the solution to the given functions and their point values in the boundary condition. The resulting inequalities impose constraints on the permissible values of the problem parameters, guaranteeing the uniqueness of the solution. Violating these constraints may lead to the problem being ill-posed or to the absence of a unique solution. The question of the existence of a solution is investigated by reducing the problem to an equivalent system of functional relations linking the traces of the desired solution and its derivative on the degeneracy line. These traces are considered separately for the hyperbolic and parabolic parts of the domain. Further investigation reduces to solving a Fredholm integral equation of the second kind. It is important to note that the kernel of this integral equation has a weak singularity, while the right-hand side is continuous. The unconditional solvability of this integral equation, and therefore the existence of a solution to the original problem, is deduced from the previously proven uniqueness of the solution. This is a significant point, demonstrating the close connection between the existence and uniqueness of a solution.

Currently, it is in demand to design methods and technologies for intelligent control of complex systems and processes that take into account situational awareness of problems, various control strategies and scenarios for achieving target situations in conditions of uncertainty. Models and methods based on fuzzy situational and fuzzy cognitive approaches take into account the specifics of situational awareness when controlling such systems and processes. The limitations of fuzzy situational models in controlling complex systems and processes under conditions of uncertainty are: the difficulty of accounting for the mutual influence of situational features due to the ambiguity of transitions from one fuzzy situation to another; the difficulty of assessing the simultaneous impact of several control decisions on various interdependent situational features; insufficient consideration of the time factor and duration of the impact of situational decisions on situational features; the difficulty of modeling scenario dynamics taking into account various strategies. Due to this, the best sequence of control decisions is formed, depending on the chosen strategy, and the time of their application is justified. The paper discusses the application of a new proposed variety of Soft Situational-Cognitive Models for intelligent control of turbocharger air installations. The results of the comparative assessment make it possible to substantiate the improvement of the quality of intelligent control and the efficiency of turbocharger air installations in conditions of uncertainty using the proposed model for various control strategies and scenarios for achieving target situations.

The current stage of development in the methodology for analyzing and synthesizing complex technical systems is characterized by a substantial increase in requirements for their design speed without a loss of functional quality. This is achieved through the application of information technologies for designing complex systems and modeling their operation processes. Within this paradigm, solving the problem of creating adaptive multi-agent systems for monitoring the technical condition of highly critical objects appears relevant. These systems would use a set of varying-accuracy measurement channels for each type of monitored parameter. This will enable the operational selection of a measuring instrument and the formation of the most rational monitoring algorithm, ensuring the required monitoring effectiveness under significant financial constraints. The aim of this work is to substantiate the possibility of using varying-accuracy measurement channels when forming the structure of each technical condition monitoring channel to ensure a given reliability. To achieve this aim, the following tasks were set and successfully solved: – modern conditions for designing technical condition monitoring systems for complex technical systems were considered; constraining factors and trends in design theory development were identified; causes of potential shortcomings in prospective systems and ways to eliminate them were determined;– the form of the system performance criterion was defined as a generalized indicator, obtained based on an additive convolution of unit unconditional monitoring reliability indicators and weight coefficients of their importance; – a structural diagram of an adaptive system was synthesized based on a model of a single-parameter monitoring channel with measurement tracts of different accuracy and the capability to implement known methods for improving monitoring effectiveness; – a computational experiment was conducted, the results of which allowed for refining the identified patterns and proposing new recommendations. The results of the conducted research enabled the development of an adaptive algorithm for monitoring a multi-parameter Object of Control based on a multi-agent system. The behavior of its agents is modeled using retrospective information from previous monitoring cycles, predictive models based on this information, as well as the implementation of the principle of equal reliability and penalty functions in case of deviation from it.

A significant drawback of classical output stages of microelectronic operational amplifiers, which today are implemented based on BJTs (bipolar junction transistors), CMOS (complementary metal-oxide-semiconductor transistors), or JFETs (junction field-effect transistors), including when these technologies are used together, is that they exhibit a relatively low output voltage slew rate under the influence of large-amplitude pulsed input signals. This undesirable effect is mainly caused by the presence of parasitic capacitances inherent in the reference current source circuits employed and in their output transistors. This paper discusses an original and effective circuit design technique that provides forced recharging of parasitic capacitances, which is used in output stages implemented in numerous patents of the world’s leading microelectronic companies. The introduction of a special differentiating transient correction circuit into the original schematics, as well as various options for its practical implementation, are considered for the first time. The circuitry of such output stages is protected by a patent of the Russian Federation. Four different modifications of buffer amplifiers have been developed and investigated, which are intended for use in modern technological processes implemented with silicon BJT or CMOS transistors, as well as with gallium arsenide n-JFET and p-n-p bipolar transistors. Examples of computer simulation of DC operating modes, as well as transient processes, are presented, which clearly demonstrate a significant increase in the maximum output voltage slew rate (by more than one to three orders of magnitude). The proposed advanced circuit solutions for GaAs transistors are recommended for practical application in microelectronic devices designed for operation under elevated temperature conditions.

The article discusses the development of a risk-oriented geoportal decision support system for territorially distributed organizational systems (TDOS). The aim of the work is to develop an architecture and a formalized model that integrates spatial data, risk structures, key performance indicators (KPIs), and management action options within a unified analytical framework. The relevance is due to the fact that classical DSS and geoportals fragmentarily cover the tasks of TDOS management due to the lack of integration of spatial analysis with risk cascading models, which leads to inconsistency of decisions and increased territorial vulnerability. The methodological foundation includes the formalization of "object-risk-indicator-impact" relationships, the construction of directed influence graphs, and mechanisms for the propagation of risk effects across the territory and management levels. A multi-layered architecture of the geoportal platform is proposed, including subsystems for spatial data collection, risk analytics, KPI dashboards, and scenario modeling. The technical implementation is based on open geoportals of the Russian Geographical Society and a unified spatial data repository. As a pilot area, the regional management system for development and natural resource use of the Republic of Mordovia was studied, where a prototype of the risk-oriented module was implemented. The article demonstrates the ability to visualize the distribution of natural and man-made risks, assess integral territorial vulnerability indices, and select priority management scenarios. The results of implementation at EM-KAT LLC showed a reduction in energy consumption. The application of the system in the Main Directorate of the Ministry of Emergency Situations for the Republic of Mordovia made it possible to optimize territorial management and increase risk predictability. The proposed approach ensures holistic and reproducible management of territorially distributed systems, increasing their resilience to local and transitive impacts, decision-making transparency, and the efficiency of interdepartmental coordination