The subject of the study is the application of spiking neural networks of the Liquid State Machine (LSM) type to the problem of chaotic time series forecasting. The objective is to establish a theoretical and software foundation for modeling the LSM utilizing the BindsNET library, alongside performing a reproducible comparative analysis against a Long Short-Term Memory (LSTM) recurrent neural network using the Mackey–Glass benchmark. The methodology includes building a reservoir of Leaky Integrate-and-Fire neurons, rate coding of the input signal via a Poisson spike generator, and training a linear readout using Ridge regression. A single-layer Long Short-Term Memory network served as the reference model. The scientific novelty of the work lies in the reproducible comparison of a spiking reservoir based on Leaky Integrate-and-Fire neurons with a Long Short-Term Memory network on a standard benchmark with open-source code. The emphasis is placed on comparing the number of trainable parameters and the practical cost of readout retraining. Main results are: the LSM achieves root mean square error (RMSE) at 0.0083 and mean absolute error (MAE) at 0.0061; readout training time is 12.4 s, full pipeline time (including reservoir simulation) is approximately 58 s. The recurrent network achieves RMSE at 0.0071 and MAE at 0.0052 with a total training time of 47.8 s. Reservoir activation sparsity is 9.3%. The optimal spectral radius ρ = 0.9 is selected by grid search over ρ ∈ {0.5; 0.7; 0.9; 1.0; 1.2} using minimum RMSE on the validation set as the selection criterion (Table 4). Statistical significance by Welch’s t-test (p < 0.05) should be considered preliminary given the limited number of runs (n = 5). The LSM underperforms the recurrent network in accuracy by no more than 17 % while training 34 times fewer parameters (501 versus 17 217). Although the full pipeline time of the LSM (~58 s) slightly exceeds the total training time of the recurrent network (47.8 s), in scenarios requiring repeated readout retraining on new data without re-simulating the reservoir, the adaptation time is reduced to 12.4 s versus 47.8 s, making the LSM a practically viable alternative when the number of trainable readout parameters does not exceed 10³ and the model re-adaptation time must not exceed 15 s on a dataset of 2 000 points.

Local area networks are widely used in modern organizations for data transmission and user interaction. The growth and development of information and communication technologies lead to the need for careful design and evaluation of the quality of functioning of local area networks before implementing new solutions. The prevalent methods employed for the calculation of bandwidth and packet delay are frequently insufficient to encompass the true operational state of a network. Such insufficiency arises from their reliance on simplifying assumptions pertaining to the nature of network load and traffic. Designing and optimizing such networks is a difficult task, especially under conditions of limited resources and increasing demands on information processing speed. This article illustrates the feasibility of employing simulation modeling techniques for the investigation of local computer networks, encompassing the assessment of their efficacy and the refinement of their architecture. A simulation model of a local computer network has been built. The settings of the main elements of the simulated process are described. Tests are conducted with the constructed model with different queue capacities and server loads. Based on the results of the tests, the parameters and characteristics of the network are clarified, conclusions on optimizing its operation and improving its functioning are noted. In particular, it was found that the network under consideration has problems with queuing. The queue size is a controllable parameter, and its optimal value can be selected experimentally using a simulation model. It is noted that improving network performance, namely reducing the percentage of lost requests, can be achieved by optimizing the queue size. The materials presented in the article serve as a basis for evaluating local computer network functionality during design and modernization, and are also relevant to the development of foundational educational programs.

The subject of this paper is a practical task: aligning the intended structure of an online course with actual user activity recorded in the standard event log of a Moodle system. The aim is to describe an original method for producing a consolidated course evaluation report and its implementation as a local plugin suitable both for a teacher’s course-level self-check and for administrative cross-course monitoring. The tasks include formalizing input data, selecting a metric set, defining access and report-storage rules, and ensuring reproducible calculations within a selected calendar interval. The methodology is engineering-oriented: the input includes course element descriptions and role- and period-filtered records from the standard log; the output includes numerical indicators for structure, activity type composition, content use, student reach, and temporal activity stability, along with a weighted aggregate score with an explicit formula and a list of low-engagement elements grouped by module type, excluding decorative labels. The result is a working extension with interfaces for running evaluations, viewing reports, deleting reports, and executing batch runs in a separate administrative mode. The conclusion confirms the practical value of the tool for teachers’ self-assessment of whether designed course elements are actually used, and for institutional monitoring, with an explicit limitation: the indicators reflect observed usage patterns and structural consistency with logs rather than learning outcomes. The novelty lies in the combined use of these metrics, role-specific access scenarios, and transparent score presentation in the report.

The article examines the influence of the semiconductor heterostructures’ configuration and shape design on the cooling performance of optoelectronic devices. The authors investigate how the heat sink geometry and the active elements’ arrangement define the heat flow paths, local overheating, and the temperature fluctuation magnitude. In terms of temperature field uniformity and heat dissipation capability, the paper compares various structural alternatives. It is confirmed that the optimal shape design and layout of the heterostructure components significantly improve the heat dissipation and the stability of the specified devices despite using the same materials.

The article presents a comparative analysis of the evolution of control methods for complex systems: from classical PID controllers to modern approaches based on Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL). This study aims to pinpoint the strengths and weaknesses of various control algorithm classes and validate where they can be effectively applied to practical tasks such as autonomous driving, industrial automation, energy, and robotics. The methodology includes a systematic review of 15 scientific sources, structured comparison based on criteria of accuracy (steady-state control error), stability (disturbances and parameter changes), and adaptability (self-tuning capability), analysis of numerical data from practical implementations. The scientific novelty lies in the development of a comprehensive methodology for comparative evaluation of classical and intelligent control systems with quantitative justification of application areas: PID provides accuracy of ±1–3% for linear objects, RL ±0.5–2% with adaptation in 10–50 iterations, DRL ±0.1–0.5% when working with high-dimensional data. The main results show that RL/DRL systems surpass PID in energy efficiency by 10–40% (confirmed by the example of Google data center cooling management with 40% savings) and adaptability in dynamic environments, but require significant computational resources (days or weeks of GPU training) and pose verifiability problems. The practical significance lies in providing control system designers with evidence-based recommendations for choosing the type of control system, considering the complexity of the controlled object, adaptability requirements, and available computational resources.

This article presents the results of measuring microclimate parameters in a residential apartment in a multi-apartment building in the northern region, focusing on carbon dioxide concentrations as the most sensitive indicator of air quality for the human body. The study examines the dynamics of carbon dioxide concentrations, temperature, relative humidity, and pressure under real-life conditions. This study aims to assess the inertia of the residential microclimate, compare measured carbon dioxide levels with current air quality guidelines, and justify the feasibility of using a digital twin to solve the problem of predictive control of a smart home system. The study methodologically uses a measuring device based on an ESP32-C3 Super Mini microcontroller with AHT20, BMP280, and SCD40 sensors, which generates time series of microclimate parameters. It is shown that during the overnight stay of one person, carbon dioxide concentrations increase monotonically from approximately 1100 cm³/m³ to values above 1800 cm³/m³, exceeding the recommended limits for residential premises. The scientific novelty of this study lies in the experimental substantiation of the fact that the bedroom microclimate can be considered an inertial system, for which the use of predictive control tools for smart home devices based on the predicted dynamics of carbon dioxide is advisable. It is also substantiated that a digital twin integrating sensor data and a room model can proactively detect deteriorating air quality and initiate the most energy-saving control actions.

The article is devoted to the concept study of performance efficiency for human-operator and efficiency rating methods by developing a prototype simulator for operator of industrial control system (ICS). Methods and algorithms that provide the simulation of production process control to assess operators’ performance efficiency rate during process control are introduced. A software tool was developed by the authors for the prototype simulation for an assessment of operator performance efficiency, which could be applicable in actual enterprises.

With the increasing complexity of modern large transportation systems, there is a growing need to integrate diverse transportation infrastructure components into a single model for their joint analysis and optimization. Hence, the paper describes representations of transportation infrastructure. The research aims to design a system model of the specified phenomenon, which would provide a comprehensive understanding of its determinants, i.e. interconnected objects, parameters, and events. The article proposes an original technique based on spatial and temporal data, as well as functional features of transportation infrastructure components. At the heart of the developed method lies a system stability index, a new measure reflecting the coherence and balance of transport network components at any given time. The work investigates the spatial and temporal interactions of transportation infrastructure objects and traffic flows. The authors implement the proposed model as part of a geographic information system on the example of Surgut and perform simulation experiments. The findings show the possible early detection of infrastructure degradation prior to major traffic congestion and disruptions. Thus, the designed model allows researchers to assess quantitatively the coherence level of the transportation infrastructure components, identify high traffic density areas, and evaluate measures to improve the state of the examined system.

The article examines the pressing problem of rapidly distributing a limited contingent of defenders (mobile agents) among a dynamically changing set of targets on a plane. In conditions where the number of objects requiring maintenance or protection is constantly changing (new targets appear or existing ones disappear), and resources are strictly limited, choosing an optimal assignment strategy becomes critical. To address this problem, the author develops a simulation model of a scanning system that monitors the spatial position of targets in real time. The proposed approach is based on the use of a greedy algorithm, which makes a locally optimal decision at each simulation step regarding the assignment of a defender to the highest-priority target. The scientific novelty of this work lies in the seamless integration of this algorithm with a dynamic cost matrix. This matrix is not static, but is recalculated for each iteration, taking into account the current coordinates of targets, their importance, and availability, allowing the system to adapt to changing circumstances. A significant contribution to the development of this topic is the creation of a comprehensive visualization system. Unlike existing analogs, it allows for the simultaneous display of not only the spatial distribution of participants (defenders and targets), but also the current state of the cost matrix, as well as visualization of the decision-making logic for assignment. The practical significance of this research extends far beyond purely defensive tasks. The developed computational method can be effectively applied in civilian sectors, such as logistics (for distributing a limited number of couriers to incoming orders), managing groups of mobile robots (organizing patrols or interactions), and in monitoring and alerting systems where rapid response to incidents is required. Thus, the work represents a ready-made prototype of an intelligent decision support system for a wide range of resource allocation problems in a non-stationary environment.

The paper addresses the flocking control problem for groups of mobile robots collectively moving in a three-dimensional environment with obstacles under limited communication between agents. The research relevance lies in the need for swarm coordination given intermittent interaction between an agent and those inside its communication range, i.e. neighbors. In this article, the goal is to create a flocking controller that adheres to collective motion principles while maintaining safety in conditions of limited visibility. The authors propose a method that integrates a learned state estimator based on a fully connected neural network, approximating the particle filter with simultaneous uncertainty evaluation, and an optimization regulator with a control barrier function and a control-Lyapunov function. The simulation reveals that the neural network estimator ensures accuracy comparable to traditional filtration techniques at a computational cost one to two orders of magnitude lower. In addition, the weighted centroid, which considers estimation reliability, reduces the possibility of coordination errors. Results can be applied in designing multi-robot group control systems.