Вісник НТУУ «КПІ». Радіотехніка, радіоапаратобудування : збірник наукових праць, Вип. 103
Постійне посилання зібрання
Переглянути
Нові надходження
Документ Відкритий доступ Analysis of a Two-Stage Thyristor Rectifier Topology with Parallel Bridges for Reactive Power Reduction(КПІ ім. Ігоря Сікорського, 2026) Komarov, V. F.; Rassokhina, Yu. V.High-power phase-controlled thyristor DC drives, despite their reliability, have a fundamental drawback: significant reactive power consumption from the supply network, especially in dynamic operating modes with large firing angles (α). This leads to a low Displacement Power Factor (DPF) and additional losses. This study analyzes a cost-effective novel two-stage thyristor rectifier topology proposed to mitigate this problem. The topology utilizes a special transformer with secondary winding taps (e.g., at 50% and 100% voltage) that feed two 6-pulse bridges (R1 and R2), connected in parallel to a common DC load. The system functions as a high-speed solid-state equivalent of an On-Load Tap Changer (OLTC). By sequentially engaging the bridges, the converter maintains small firing angles (α) over a wide output voltage range. A detailed analysis of the energy characteristics, based on a case study of a typical acceleration cycle for a high-inertia drive (hoisting machine), demonstrates the topology's energy efficiency. Compared to a conventional single-bridge thyristor rectifier, the two-stage scheme reduces the total reactive energy consumed per acceleration cycle by 51% (from 54.4 kVAr · h to 26.8 kVAr · h in the example). An additional advantage of the solution is the reduction of the Root-Mean-Square (RMS) primary winding current during the initial acceleration stage, which leads to lower active power (I2R) losses in the transformer and supply lines. A drawback of the considered commutation algorithm is the presence of a 'discontinuous current mode' in the transformer primary winding during the mixed-mode (simultaneous operation of R1 and R2), which significantly degrades the harmonic spectrum (THD) of the input current, making the converter non-compliant with power quality standards (e.g., IEEE 519). The study concludes that practically implementing this energy-efficient topology requires hardware adaptation, specifically through the design of a custom Inter-Phase Reactor (IPR) or integration into multi-pulse configurations, to mitigate harmonic distortion while preserving the reactive power benefits.Документ Відкритий доступ Model of Infrared Radiation Polarization of a Drone(КПІ ім. Ігоря Сікорського, 2026) Kolobrodov, V. G.A physico-mathematical model of infrared (IR) radiation polarization from a drone, which can be used in the creation of polarimetric thermal imagers (PTI) for detection and recognition of surveillance objects, is developed. Analysis of Research. Thermal imaging systems are widely used, primarily in military applications, such as in the thermal cameras of drones. The principle of operation of classical thermal imagers is based on converting the brightness of IR radiation from the observed object and background from the plane of objects into an appropriate distribution of background-target scene (BTS) brightness on a display screen within the visible spectrum. If the contrast is low or absent, it becomes impossible to detect such an object. Current models do not simultaneously account for the drone’s emission and reflected radiation, leading to errors in the detection range estimation. The use of polarization properties of radiation allows solving this problem. Therefore, developing and researching a model of IR radiation polarization from the drone is a very important task for creating promising PTI for drone detection. Unresolved parts of the overall problem include the lack of a model that considers both emission and reflected polarization simultaneously. The purpose of the article is to develop a physico-mathematical model of IR radiation polarization from a drone, which can be used in the creation of PTI systems designed for drone detection and recognition. Research material presentation — the drone is modeled as a flat plate characterized by an reflection coefficient and a complex refractive index, which allowed the development of methods for calculating parameters of elliptically polarized radiation. Analysis of the developed methods indicates that for modeling the polarization state of the observed object’s radiation, it is appropriate to select the image intensity, degree of polarization, and polarization angle, all determined by the Stokes parameters. Comparison of obtained results with those from other researchers involves scientifically substantiating that the polarization of IR radiation from objects and backgrounds, due to their own and reflected radiation, has an opposite character, which significantly worsens the overall degree of polarization. Conclusions from this research. The obtained results are relevant for developing a test object model, necessary for designing PTI systems. Future prospects include conducting experimental measurements of IR radiation polarization from real drones and atmospheric conditions, which will help refine the test object parameters.Документ Відкритий доступ Infrastructure for the Real-Time Biosignal Datasets Collection(КПІ ім. Ігоря Сікорського, 2026) Mosiichuk, V. S.; Sharpan, O. B.; Yalosovetskyi, V. I.The effectiveness of modern methods of biosignal analysis largely depends on the availability of structured data sets containing both primary signals and related metadata. At the same time, most existing biosignal registration systems do not provide automated accumulation of such data in centralized databases, which complicates the formation of training samples and the further application of machine learning algorithms. The article considers an automated system for forming representative sets of biomedical data to increase the efficiency of machine learning tasks in the analysis of the human physiological state. The infrastructure of the information system for automated collection of biosignals and metadata and formation of data sets in real time has been developed. The system has a multi-level architecture that includes a sensor level for registering biosignals, a communication level for data transmission, and a server level for their processing and storage. Signals were registered during system testing using a photoplethysmographic sensor integrated with a wireless module based on a microcontroller with Wi-Fi support. The server part is implemented in a virtualized environment using open source software, a web server, a database management system, and signal processing software modules. For visualization of biosignals and interaction with users, a Web-API and a web interface have been developed, which provide access to measurements, metadata management, and signal visualization. The system implements a data streaming pipeline that includes query verification, signal storage, and biomedical parameter calculation to assess the user’s functional state. An experimental study of the system’s performance was conducted in a load testing mode that simulates the simultaneous operation of a significant number of sensor devices. The results showed that the developed infrastructure is capable of handling more than a hundred simultaneous connections with an average query processing time of less than 100 ms. The results obtained confirm the possibility of using the proposed system for scalable collection of biosignals and the formation of data sets suitable for further application of machine learning methods.Документ Відкритий доступ Computational Approaches for Emotional Burnout Detection: Machine Learning and Deep Learning Evaluation(КПІ ім. Ігоря Сікорського, 2026) Mushta, S. A.; Mushta, I. A.; Popov, A. O.; Lysenko, O. M.; Tukaiev, S. V.This study focuses on the development and evaluation of computational techniques for classifying emotional burnout based on quantitative data analysis. Instead of relying on subjective psychometric questionnaires, we investigate the applicability of machine learning and deep learning algorithms to structured datasets. Several classical methods, including logistic regression, random forest, and gradient boosting, were systematically compared with a Deep Learning (DL) ensemble model. To enhance robustness, preprocessing steps such as feature selection, data balancing, and resampling were applied. The deep learning architecture, incorporating focal loss and adaptive threshold optimization, achieved the best performance. On 5-fold cross-validation, the proposed DL model obtained an overall accuracy of 86.3%, with precision of 0.815/0.887, recall of 0.786/0.904, and F1-scores of 0.800/0.895 for the negative and positive classes respectively. The results demonstrate that advanced computational models can provide scalable and generalizable tools for automatic detection tasks, forming a technical foundation for future integration into applied research and occupational health monitoring systems.Документ Відкритий доступ Improving EMG Signal Classification with Transfer Learning Under Low-Data and Cross-Subject Conditions(КПІ ім. Ігоря Сікорського, 2026) Kolomiiets, B. Yu.; Karpliuk, Ye. S.Surface electromyography is a non-invasive method used for monitoring muscle activity and is widely applied in rehabilitation, prosthetics, assistive robotics, and human–computer interaction. However, its practical use often remains limited by large differences between individuals and the effort required to train models for each new user. This study explores whether transfer learning can help address these challenges when using deep learning to classify hand and wrist gestures. The experiments use a dataset that includes eleven gestures, each repeated eight times by 22 healthy participants. Three training approaches are evaluated: (i) training and testing on the same subject (intra-subject), (ii) training on some subjects and testing on a new one (inter-subject), and (iii) transfer learning with and without resetting the fully connected output layer of the convolutional neural network. All models are evaluated using a leave-one-out cross-validation strategy across both subjects and repetitions. Results show that both transfer learning methods outperform the other two approaches in terms of classification accuracy. The best performance is observed when the fully connected layer is reset before fine-tuning (F1-score = 0.907, σ = 0.074). Wilcoxon signed-rank statistical tests confirm that these improvements are statistically significant, even when only a few repetitions are used for fine-tuning. In fact, using transfer learning with just four repetitions instead of eight achieves accuracy comparable to training from scratch on all eight repetitions. These findings suggest that fine-tuning pre-trained models can significantly reduce the effort needed to adapt EMG-based systems to new users, providing a practical and effective approach for developing user-friendly interfaces suited to assistive and rehabilitation applications.Документ Відкритий доступ Multitarget Tracking Algorithm With Joint Probabilistic Data Association Using Coordinate and Amplitude Information(КПІ ім. Ігоря Сікорського, 2026) Kovtun, I. S.; Zhuk, S. Ya.The widespread use of small Unmanned Aerial Vehicles (UAVs) makes the task of their tracking highly relevant, especially in conditions where objects are at close ranges and their trajectories intersect. Frequency-Modulated Continuous-Wave (FMCW) radar is a modern tool for detecting and tracking small UAVs, allowing for a significant reduction in peak transmit power, thus lowering energy consumption and improving the weight, size, and cost characteristics of the system. Small UAVs have extremely low radar cross-section values. Increasing the detection range of small UAVs by FMCW radar can be achieved by lowering the detection threshold, which, however, leads to a significant increase in the probability of false alarms. To improve the efficiency of multitarget tracking using FMCW radar data in the presence of a significant number of false alarms, the Amplitude-Aided Joint Probabilistic Data Association Filter (AA-JPDAF) algorithm has been developed. This algorithm proposes the use of decision statistics (amplitude information) from the output of the optimal primary signal processing receiver as additional information. This information is utilised at the data association stage based on the Joint Probabilistic Data Association (JPDA) method. Target motion parameter estimation for each trajectory is performed using the Extended Kalman Filter (EKF). The analysis of the AA-JPDAF algorithm and its comparison with the conventional JPDAF were conducted via statistical simulation for scenarios involving intersecting trajectories and long-term parallel motion of targets at close ranges.Документ Відкритий доступ Methods for Calculating the Radio Jamming Zones of Jamming Transmitters Against the Receivers of Unmanned Aerial Vehicle Control Radio Links(КПІ ім. Ігоря Сікорського, 2026) Nahorniuk, O. A.The experience of the Russian-Ukrainian war shows that an effective way to counteract unmanned aerial vehicles is by jamming the receivers of their radio control channels. For this purpose, short-range electronic warfare systems are widely used in the tactical level. During the formulation of requirements for such systems, as well as the planning of their intended use, there is a need to calculate their effective zones of action. A mathematical model for the process of radio jamming unmanned aerial vehicles control channels receivers, which takes into account the technical parameters of the radio channel and the jamming transmitter, the heights of their antennas, and their radiation patterns are developed in the article. It is shown that the radio jamming zone of an interference transmitter with an omnidirectional antenna is limited by a sphere in space, and by circles on the ground surface, the centers of which are shifted relative to the location of the jamming system. The use of directional antennas makes it possible to increase the range of the jamming transmitter, but the jamming zone will be sectoral in space, and its shape will depend on the antenna's radiation pattern. Four methods for calculating the radio jamming zones of jamming transmitters against unmanned aerial vehicles control channel receivers, corresponding to the most common practical cases of their application, are proposed in the article. Software for calculating and graphically displaying the jamming zones has been developed in the MATLAB programming environment. Examples of using the proposed methods for calculating the radio jamming zones of individual protection active jamming systems with ''dome-type'' antennas and group protection systems with directional antennas are given.Документ Відкритий доступ Optimization of Laser Hazard Warning Systems(КПІ ім. Ігоря Сікорського, 2026) Asadov, H. H.; Guliyev, F. F.; Nematzade, R. G.Actuality. Active and semi-active optical guidance systems for projectiles and missiles at a target mainly use a laser beam from a Laser danger Warning System (LWS) designed for prompt notification of the fall of a laser targeting beam on an important protected object. Photodetectors are used to register the laser beam hitting the surface of the protected object. To reduce the likelihood of a false alarm, it is necessary to achieve the possibility of receiving the largest possible optical signal on photodetectors. Setting the task. To increase the accuracy of the generated alarm signal based on the signals of photodetectors detecting direct laser beam hits, it is proposed to install additional remote photodetectors for recording reflected and diffusely reflected optical signals. Method. The problem of optimal selection of the ratios of distances from the laser to the target and from the target to additional receivers in LWS with remote photodetectors has been formulated and solved. Two subtasks are formulated: in the first, the target indicated by the laser belongs to the enemy, and in the second, the opposite side. It is shown that when optimizing systems of laser-guided missiles and projectiles, taking into account the LWS introduced by the enemy, it is possible to use the theory of the Laser Detection and Ranging (LADAR) systems. The result is that the application of the theory of radar systems has made it possible to optimize the functioning of the attacking side in terms of optimal choice of the distance to the enemy using the LWS with remote photo sensors.Документ Відкритий доступ Throughput Capacity of RF Sensor for Unmanned Aerial Vehicle(КПІ ім. Ігоря Сікорського, 2026) Buhaiov, M. V.Usage of small unmanned aerial vehicles (UAVs) for spectrum sensing, especially in urban areas, has numerous advantages over the use of ground-based stations for radio frequency (RF) emitters detection and location. In order to develop spectrum sensing equipment for UAVs, it is necessary to establish a number of requirements for it. One of the main requirements is the necessary throughput capacity. The purpose of the article is to improve the methodological apparatus to establish requirements for UAVs spectrum sensing equipment. To describe the density of RF emitters distribution, it is proposed to use a nonhomogeneous Poisson spatial process in combination with parametric or nonparametric distribution functions. The density function of this distribution reflects the average number of RF emitters that are within energy accessibility and can be detected. Using a quantile of a given Poisson distribution level, in which the density function is used as a parameter, allows to estimate maximum number of RF emitters. The signal flow from each RF emitter is described using a nonstationary Poisson process. The moments of time of broadcast and the duration of signal emission are exponentially distributed. Estimates of the average intensity of RF emitters during analyzed time interval of a given frequency band for a single-channel multi-antenna system have been obtained. The methodology for estimating the required throughput capacity of RF sensor and recommendations for using the proposed methodological apparatus in conditions of a priori uncertainty regarding the density of RF emitters distribution and signal flow intensity are presented. Using the values of the maximum number of RF emitters within the energy availability range for the entire spectrum sensing area, the average intensity of RF emitters, and the analysis time of the instantaneous frequency band, it was obtained an estimate of required throughput capacity of RF sensor.Документ Відкритий доступ Microwave Transistor Oscillators Development Based on Dielectric and Microstrip Resonators(КПІ ім. Ігоря Сікорського, 2026) Avdeyenko, G. L.; Zhivkov, O. P.; Ilchenko, M. Yu.; Krylach, O. F.; Stepanenko, V. M.The paper shows the possibility of creating microwave range transistor oscillators in Ukraine based on the use of both a Dielectric Resonator (DR) and a Microstrip Metamaterial Resonator (MMR) in the form of a Split-Ring Resonator (SRR), with output oscillation parameters that are comparable to the output oscillation parameters of similar transistor oscillators currently developed by leading scientific groups in the world. The theoretical part of the work demonstrates analytical and graphical characteristics of the S11 reflection coefficients of the proposed MMR in the form of an SRR and justifies the realistically achievable level of improvement in the quality factor of a resonant system consisting of two SRRs compared to the quality factors of individual SRRs. Using the Signal Hound and Rohde & Schwarz spectrum analyzers, the output oscillation parameters of four samples of the developed transistor oscillators in the 4 GHz and 10 GHz ranges were investigated, three of which are built on the basis of a HEMT Field-Effect Transistor (FET), and the fourth — on a bipolar transistor. The output frequency of two transistor oscillators — on a FET and bipolar transistor stabilized by a DR. The output frequency of the other two transistor oscillators is stabilized by an MMR in the form of single and double SRRs respectively. Comparison of the data obtained from the measurement results of the spectral density of phase noise for offset frequencies of 10 kHz showed that the developed transistor oscillators on a bipolar and HEMT FET stabilized by a DR are equivalent to foreign analogues of such oscillators. In turn, for offset frequencies of 100 kHz and 1000 kHz, the developed transistor oscillators stabilized by an MMR equivalent to foreign transistor oscillators stabilized by an MMR with more complex geometric configuration. The paper may be useful for developers of microwave infocommunications devices, in particular radio-relay, tropospheric and satellite communication systems.Документ Відкритий доступ Electromagnetic Analysis of an Inductive Iris in a Rectangular Waveguide via a Hybrid Mode-Matching and Integral Equation Technique(КПІ ім. Ігоря Сікорського, 2026) Piltyay, S. I.Inductive irises are applied in modern microwave filters, diplexers, polarizers, and rotators based on waveguides. The article presents an efficient mathematical technique for the analysis of the characteristics of the electromagnetic waves scattered by an inductive iris in a rectangular waveguide. Using the mode-matching technique for the transverse field components, the electromagnetic waves scattering problem was reduced to a set of coupled integral equations, which were then decoupled. Each equation was solved by expanding the electric field in the iris window into a series of basis functions. This procedure was implemented using sets of orthogonal trigonometric basis functions of the aperture field, orthogonal basis functions based on Gegenbauer polynomials with a weighting function of power 1/2, or orthogonal basis functions based on Gegenbauer polynomials with a weighting function of power 2/3. As a result, it became possible to determine the phasors of all modes in each region of the inner volume (before the iris, inside its aperture, and after the iris), along with the complex reflection and transmission coefficients of the fundamental electromagnetic mode TE10. In order to validate the correctness and accuracy of the developed mathematical model, additional calculations by efficient numerical methods (Finite Element Method and Finite-Difference Time-Domain method) and experimental measurements of reflection characteristics were performed for two inductive irises in a standard rectangular waveguide. Measurement setups included a scalar network analyzer or a vector network analyzer, one of two inductive irises, a matching load, and waveguide channels. The experimentally obtained reflection coefficients were in good agreement with those predicted by the developed mathematical model and numerical methods. The developed mathematical model can be widely applied to the analysis of electromagnetic wave scattering by inductive irises in waveguides and to the synthesis of various microwave devices based on these irises.