2687-0517 18 1 2025 1-140

RAR RUS 7-22 0000-0002-1562-2676 Ageev Andrey F-6480-2013 7004013271 0000-0002-5637-1420 Peter the Great St.Petersburg Polytechnic University Lev V. lev.utkin@mail.ru

Polytechnicheskaya, 29, St.Petersburg, Russia, 195251

Peter the Great St. Petersburg Polytechnic University Konstantinov Andrei andrue.konst@gmail.com

St. Petersburg, Russian Federation

Improved anomaly detection by using the attention-based isolation forest with trainable scoring function This paper proposes a novel anomaly detection model, called Attention-Based Isolation Forest with trainable Scoring Function (ABIF-SF). ABIF-SF enhances the original isolation forest algorithm by incorporating attention weights determined by scoring functions whose parameters are trained using gradient descent. The attention weights indicate the relevance of each data instance to the anomaly assessment task for each tree in the isolation forest. Two scoring functions are explored – scaled dot product and additive attention. Numerical experiments on real-world datasets demonstrate that ABIF-SF achieves better anomaly detection performance compared to isolation forest and attention-based isolation forest with the contamination model. The proposed method simplifies the computation of attention weights by using scoring functions and hinge loss optimization. The code implementation of ABIF-SF has been made publicly available for further research and benchmarking. Overall, the incorporation of trainable scoring functions to compute context-aware attention weights improves isolation forests for anomaly detection tasks. 10.18721/JCSTCS.18101 004.85 anomaly detection attention mechanism isolation forest Nadaraya–Watson regression quadratic programming contamination model additive attention https://infocom.spbstu.ru/article/2025.84.1/

RAR RUS 23-35 Sabutkevich Artem artem.sabut@gmail.com Nikiforov Igor igor.nikiforov@gmail.com Peter the Great St.Petersburg Polytechnic University Samochadin Aleksandr Samochadin@soft-consult.ru

Polytechnicheskaya, 29, St.Petersburg, 195251, Russia

A method for modeling of individual agent behavior in the process-network paradigm of discrete-event simulation Due to the active growth of demand for cloud resources, the task of increasing the efficiency of their use becomes relevant. One of the approaches to solving this problem is the application of discrete-event simulation in the process-network paradigm, which allows describing the modeled process in the form of processing nodes united in a single network. However, this paradigm does not consider the individual behavior of agents, which reduces the adequacy of the resulting models. The paper presents an approach to assessing the adequacy and significance of simulation models, and proposes a method that allows specifying and considering of the individual behavior of agents in the simulation process, implemented in accordance with the process-network paradigm. The application of this method allows increasing the applied significance of the models of the processes under study. The paper describes the integration of the proposed method into systems implementing the process-network paradigm and presents its software implementation. The latter allows to investigate the influence of considering individual agent behavior on the adequacy and significance of the model of virtual machines placement relative to physical servers. Due to the application of the method, it was possible to achieve an increase in the adequacy of the model under study by an average of 12.5% and, as a consequence, to increase the number of significant models by 65% for the selected adequacy threshold value. 10.18721/JCSTCS.18102 004.94 discrete-event simulation agent-based simulation process-network paradigm model adequacy simulation of cloud infrastructures https://infocom.spbstu.ru/article/2025.84.2/

RAR RUS 36-47 Alekseev Evgeny evgeny.alekseev@uec-saturn.ru 57189242005 0000-0001-9271-1552 Lomanov Alexey frei@rsatu.ru Ivanov Dmitry dmitry.ivanov@uec-saturn.ru A technique for automated analysis of the blade surface for defects under UV light A technique for automated analysis of the blade surface for defects under UV light is presented. The basis of control operations when inspecting blade surfaces is the use of machine vision. The technique solves several key problems: obtaining a package of inspection images of a complex profile object of inspection (an aircraft blade), determining the actual parameters (sizes) of glows for single and group defects, generating expert recommendations (digital trace) for determining the presence of defects on the inspected surfaces for the operator or automated systems. An algorithm for processing images obtained from a video camera is presented, and approaches to compensating for the shift of blades in a frame during inspection rotation are described. The technique describes the following sequentially performed stages: shooting of the blade surface; searching for glows in a two-dimensional image; converting two-dimensional coordinates of the glows into three-dimensional ones; determining the actual parameters of the glows; determining the position of the glows relative to each other; determining the degree of suitability of the blade based on the obtained information about the glows. 10.18721/JCSTCS.18103 681.5.08 non-destructive testing machine vision technological process automation measurement systems recommendation system https://infocom.spbstu.ru/article/2025.84.3/

RAR RUS 48-59 0009-0009-1102-7333 Peter the Great St. Petersburg Polytechnic University Shulgin Sergey shulginsergey0@gmail.com

St. Petersburg, Russian Federation

Peter the Great St. Petersburg Polytechnic University Benderskaya Elena helen.bend@gmail.com

St. Petersburg, Russian Federation

Leveraging natural language processing techniques for enhanced recommender systems Recommender systems often use NLP methods primarily for content processing. In this study, we propose a new approach to building recommender systems, in which user interaction data with content is considered within the framework of a natural language model. Thus, the user preference vectorization model Pref2Vec is proposed as the basis for a hybrid recommender system. Moreover, a concept of a user embedding space (UES) is introduced, which represents a set of extended embeddings that capture end-user preferences. A new method of applying clustering analysis to the recommendation process is also proposed. The Pref2Vec model and the UES class were implemented in the Python programming language as an extension of the functionality of the Gensim library. The model was evaluated using Recall@k and NDCG@k metrics. The comparative analysis showed that the results obtained are comparable with the performance of the BPRMF, GRU4Rec and NextItRec models, which indicates the potential of the proposed model. 10.18721/JCSTCS.18104 004.85 recommender system natural language processing NLP methods cluster analysis 2Vec models vectorization of user preferences embedding https://infocom.spbstu.ru/article/2025.84.4/

RAR RUS 60-71 0000-0002-2813-2676 Peter the Great St. Petersburg Polytechnic University Aliyev Ali aliev.aa@edu.spbstu.ru

St. Petersburg, Russian Federation

Molodyakov Sergey sm50@yandex.ru ResNet-SV: Fast and accurate speaker verification with a multi-layer cascade attention mechanism One of the most challenging issues of voice biometrics rapid development is the need to develop methods that can combine speed and accuracy. Traditional solutions tend to choose a compromise between these two aspects, which either complicates the speaker verification process or reduces accuracy, especially under real-world conditions in which background noise and fluctuation in speech are substantial obstacles. This paper examines modern approaches and their architectural features. The architecture is based on ResNet, originally designed for computer vision tasks, which was modified and adapted for optimal performance in speech processing. The proposed modification method based on a multi-layer cascade attention mechanism for feature extraction from convolutional blocks is described in detail. This modification allows using fewer layers for feature extraction, thereby increasing the speed of the model, and allows to deal more effectively with the noise in the audio signal. The paper concludes with the model parameters used in the training process, as well as key metrics such as EER and minDCF computed on the VoxCeleb1 dataset. The results are compared with solutions built on other architectures. Through experimentation, the authors were able to achieve a high level of accuracy, with a smaller number of the neural network model parameters. This work brings us closer to a wider application of voice biometric systems in various scenarios. 10.18721/JCSTCS.18105 004.89 speaker verification speaker identification voice biometrics convolutional neural networks attention mechanism speech processing https://infocom.spbstu.ru/article/2025.84.5/

RAR RUS 72-84 0009-0004-1628-1772 Peter the Great St. Petersburg Polytechnic University Pham Huu Duc phamduc2511997@gmail.com

St. Petersburg, Russian Federation

22735712200 Peter the Great St. Petersburg Polytechnic University Sorotsky Vladimir sorotsky@mail.spbstu.ru

St. Petersburg, Russian Federation

Characteristics of Class E power amplifier with complex impedance load Unlike the well-known publications focused on the analysis of the characteristics of a Class E power amplifier (PA), in which the authors limit themselves to considering a particular case of a real load, this paper presents the results of calculating the characteristics of a Class E PA with a complex impedance load. It is especially relevant when operating in a frequency band or amplifying broadband signals. The relations given in this paper can be used to solve two types of problems. In the first case, related to “soft-switching” mode Class E PAs characteristics can be determined, when the voltage on the transistor and its derivative at the moment of turn-on are equal to zero, which eliminates switching losses. In the second case, the problem of synthesizing a matching circuit that ensures the operation of the PA in the extended frequency band can be solved. The matching circuit synthesis can be carried out under the limitations on the acceptable change of output power and voltage drop on transistor just before switching. 10.18721/JCSTCS.18106 621.37 power amplifier efficiency Class E complex impedance load analytical model simulation modeling harmonic balance https://infocom.spbstu.ru/article/2025.84.6/

RAR RUS 85-97 Ivanov Nikita ivanovnick@mail.ru Rumyancev Ivan i.a.rumyancev@gmail.com Monolithic microwave bandpass filters design for S- and C- bands 10.18721/JCSTCS.18107 621.372.543 MMIC design bandpass filters S-band C-band GaAs pHEMT https://infocom.spbstu.ru/article/2025.84.7/

RAR RUS 98-110 Concern CSRI Elektropribor, JSC Tulaev Artyom artulaev@gmail.com

St. Petersburg, Russian Federation

Concern CSRI Elektropribor, JSC Kozlov Alexey kas573@yandex.ru

St. Petersburg, Russian Federation

0000-0003-4379-5803 Kostygov Dmitrii 0009-0003-3140-9896 Kuznecov Kirill Belyaev Yakov designcenter.spb@mail.ru 0000-0003-3103-7060 Peter the Great St.Petersburg Polytechnic University Loboda Vera

Polytechnicheskaya, 29, St.Petersburg, 195251, Russian Federation

A micromechanical pressure sensor with reconfigurable ASIC In this paper, a MEMS pressure sensor with reconfigurable ASIC and measurement range of 10 kPa to 100 MPa was designed, based on the system model design approach. The competitive advantages of the method include rapid prototyping and controlling parameters of micromechanical pressure sensor at all design stages; thus, reducing the device development and manufacturing costs. The set of unified piezoresistive sensing elements with different full scale pressure and sensitivity was implemented on a pre-doped silicon on insulator (SOI) wafer. The optimal parameters of sensing elements and integrated circuit were obtained using complex optimization criterion by system level simulation and refined by finite element method. The reconfiguration requirements were obtained by simulation of technological process variations. The reconfigurable ASIC is implemented using 0.18 µm SOI technology. The ASIC provides integrated solution with on-chip programmable offset trimming, temperature sensing, clock generation and digital signal processing. The system level and schematic simulations were performed during ASIC development. The digital signal processing verification was performed by FPGA prototyping. The experimental studies were carried out for sensor prototypes with 100 kPa and 1 MPa full scale range. The developed pressure sensor based on micro-assembly achieves the 0.06% main full-scale error. 10.18721/JCSTCS.18108 681.586.2, 621.3.049.7 MEMS ASIC SOI pressure sensor system model reconfiguration https://infocom.spbstu.ru/article/2025.84.8/

RAR RUS 111-129 0000-0001-6425-9749 Peter the Great St. Petersburg Polytechnic University Khrustaleva Irina irina.khrustaleva@mail.ru

St. Petersburg, Russian Federation

AAH-8784-2019 35303230700 Peter the Great St.Petersburg Polytechnic University Vyacheslav P. shkodyrev@imop.spbstu.ru

Polytechnicheskaya, 29, St.Petersburg, 195251, Russia

Khokhlovskiy Vladimir 78v.kh77@gmail.com Chernyh Larisa 2904180@mail.ru Stepanov Sergey stepanov56@mail.ru Optimization model of the processing parameters for structural elements of a product Optimization of the target indicators of the technological process is a key factor in increasing the efficiency of the product manufacturing process. The efficiency of the optimization process directly depends on the degree of detail of the control object. The purpose of the study is to increase the efficiency of the process of forming individual geometric elements of a part through multi-criteria optimization of the technological process parameters. The paper presents a structural hierarchical model of optimizing the parameters of the process of forming a geometric element. This model is a structural decomposition of the goals to be achieved within the identified control level. Based on the structural decomposition, four levels of process control are identified. This hierarchy of goals allows increasing the efficiency of the geometric element formation process through detailed analysis and optimization of target indicators at each stage of the process. The paper considers an example of optimization of the process parameters for machining a group of threaded holes M27x2-6H in a product made of dispersion-hardened composite alloy SAS-50. Optimum values of the process parameters for each forming stage are determined for the investigated group of holes according to the structural model of the process. As a result of optimizing the process parameters, the accuracy of manufacturing a group of threaded holes increased by 22.2%, while the labor intensity increased by 13.69%. 10.18721/JCSTCS.18109 62-519 multi-criteria optimization geometric element processing route structural hierarchical model control level https://infocom.spbstu.ru/article/2025.84.9/

RAR RUS 130-140 0009-0000-2651-899X Saint Petersburg Mining University Novak Diana novak_da@pers.spmi.ru

St. Petersburg, Russian Federation

0009-0002-7544-9155 Nushtaev Nikita 0009-0006-1822-7117 Saint Petersburg Mining University Kozhubaev Yury kozhubaev_yun@spbstu.ru

St. Petersburg, Russian Federation

Technological process control of oil-based gas absorption The main purpose of the paper is to study the optimization of the technological process of oil gas absorption. For this purpose, a complete analysis of the technological process was made with the identification of automation tasks: ensuring a stable temperature of absorbent in the circuit; ensuring a stable temperature of the cooling circuit; filtration of absorbent; ensuring a stable gas pressure in the system; free flow of absorbent between the tanks; accounting of purified gas. The process of selecting equipment for development of a three-level automated control system for oil gas absorption was investigated. The system has 35 discrete signals and 17 analog signals. The measuring devices that should be responsible for collecting and transmitting process information to the logic controller model were selected, the actuators that directly interact with the gas absorption process were selected. Based on the selected sensors, the type of sensor signal, its name and the required number for the possible realization of the automated system were specified. The industrial logic controller, which meets all the requirements of the technological process, was selected. The article provides a rationale for the choice made. 10.18721/JCSTCS.18110 622.279.8 oil gas absorption resource processing technological process automated control system SCADA system https://infocom.spbstu.ru/article/2025.84.10/