ISSN 0236-235X (P)
ISSN 2311-2735 (E)

Journal influence

Higher Attestation Commission (VAK) - К1 quartile
Russian Science Citation Index (RSCI)

Bookmark

Next issue

2
Publication date:
16 June 2024

Journal articles №1 2024

11. Modifying the algorithm for frontal modeling of accidental release consequences based on the empirical and statistical approach [№1 за 2024 год]
Authors: L.O. Chernyshev (plumber63@mail.ru) - Tver State Technical University (Postgraduate Student); Matveev, Yu.N. (matveev4700@mail.ru) - Tver State Technical University (Professor), Ph.D;
Abstract: The paper examines an empirical-statistical approach to constructing a cellular model for visualizing the consequences of toxic substances being released into the atmosphere in case of a local accident. It is shown that under conditions of a priori uncertainty of input data, the computing core of the supervisory system, which reproduces accidental release consequences, should implement a two-loop information processing scheme. The effectiveness of parametric estimation of a model in the outer loop of such scheme largely depends on the speed and accuracy of model computations implemented by algorithms in the inner loop of modeling and visualizing release consequences. The paper analyzes features of parametric estimation procedures under conditions of information deficit for a continued release of toxic substances. It forms requirements for an alternative modification of the modeling algorithm taking into account the advantages of the empirical-statistical approach. There is a brief description of the developed algorithm: the choice of a sixteen-point modeling template is substantiated; the features of an empirical function modifying a field of distances depending on the angular direction of wind mass transfer are considered; a detailed algorithm block diagram and the main relations forming the distance map calculation basis with the subsequent assessment of the upper limits of pollutant concentration are disclosed. The paper identifies the advantages and disadvantages of the practical implementation of the algorithm. There are results of testing the algorithm while processing actual experimental data on a conditional model of a terrain map. The proposed approach will increase the performance of the frontal modeling algorithm and reduce the time spent on searching for a reference solution in a two-loop information processing scheme. The paper materials can be used to improve the functionality of supervisory decision support systems in eliminating the consequences of accidental releases.
Keywords: computer modelling, operational response, frontal modeling, empirical-statistical approach, distance field, render
Visitors: 102

12. Risk-oriented approach to designing an anti-terrorist protection system of educational institutions [№1 за 2024 год]
Authors: Vladimir M. Kolodkin (kolodkin@rintd.ru) - Udmurt State University (Professor), Ph.D; Dina M. Varlamova (dina@rintd.ru) - Udmurt State University (Senior Lecturer); Artem D. Shakirov (artdmshakirov@gmail.com) - Udmurt State University (Graduate Student), Undergraduate;
Abstract: The paper presents a problem-oriented software package for computer forecasting of the consequences of terrorist attacks on educational institutions. One of the main forecasting tools is mathematical modeling. The complex simulates the development dynamics of an antagonistic conflict between a violator (terrorist) and risk recipients in a educational institution. When building a software package, the authors used a new concept of countering a terrorist attack. The concept feature is countering a terrorist attack by risk recipients expressed in moving human flows into security zones along safe trajectories in a building. The software package supports data integration of the spatial information model of a building, the intruder model characteristics, the characteristics of the controlled movement of human flows in emergency situations. The spatial information model of a building is created in a domestic BIM system Renga. The topological graph corresponding to the building topological model is constructed by a specialized plugin created by Renga. The advantage of the author's software package is the automatic mode of designing the reaction of risk recipients to violator’s actions. The mode provides damage minimization. Designing takes into account the characteristics of an institution engineering and technical protection system. The software package originality is to ensure the process of designing safe ways for moving human masses in a real-time emergency situation development. Real-time mode support provides a fundamental opportunity to build a decision support system based on a problem-oriented software package. The practical significance of the complex is also due to the possibility of using it as a simulator for training persons responsible for the integrated safety of educational institutions in emergency situations. The paper shows the application of the software package for ranking educational institutions by the anti-terrorist security level.
Keywords: software package, numerical simulation, anti-terrorist security, risk-oriented approach, educational institutions
Visitors: 98

13. Modeling biotechnological processes using a mathematical apparatus of artificial neural networks [№1 за 2024 год]
Authors: Sergey P. Dudarov (dudarov.s.p@muctr.ru) - D. Mendeleev University of Chemical Technology of Russian Federation (Associate Professor, Dean), Ph.D; Ilya V. Maklyaev (makliaev.i.v@muctr.ru) - D. Mendeleev University of Chemical Technology of Russian Federation (Student, Assistant), ; Yury A. Lemetyuynen (yurylemet@gmail.com) - D. Mendeleev University of Chemical Technology of Russian Federation (Student); Guseva E.V. (eguseva@rally-online.ru) - D. Mendeleev University of Chemical Technology of Russian Federation (Associate Professor), Ph.D; Boris A. Karetkin (karetkin.b.a@muctr.ru) - D. Mendeleev University of Chemical Technology of Russian Federation (Associate Professor), Ph.D; Svetlana A. Evdokimova (evdokimova.s.a@muctr.ru) - D. Mendeleev University of Chemical Technology of Russian Federation (Assistant);
Abstract: The paper focuses on studying and applying neural network technologies and tools for mathematical modeling and computer analysis in terms of biotechnological processes. The research considers modeling processes associated with gut microbiota functioning meaning microorganisms residing in the human intestine and performing several crucial functions for his health. To gather necessary data and construct models, the authors collected various indicators through experiments using a fermenter. These studies were conducted under various initial conditions: different concentrations of micro-organisms and nutrient substrate, with different environmental components. The acquired data became a base for two da-tasets: training and testing. The neural network modeling method was chosen as the research approach. Based on the training and testing datasets, neural network models were trained and subsequently tested for accuracy. A two-layer perceptron was employed as a neural network structure. The research resulted in special software to facilitate neural network modeling of biotechnological processes and to provide a mathematical description of the metabolic processes of bifidobacteria. This software was used to study relationship between the initial conditions, fermentation conditions, and bifidobacteria metabolism. The modeling results were analyzed and compared with alternative methods; they confirmed their high efficiency and the feasibility of using the neural network approach for modeling biotechnological processes. It was corroborated that using neural network models is a promising direction in the discussed domain. Due to their versatility and learning capability, neural networks can be effectively used for analyzing and describing complex processes, particularly the metabolism of gut microbiota. The developed software and algorithmic solutions offer models characterized by high accuracy and reliability. Consequently, they can be used for devising new methods for monitoring and optimizing biotech-nological processes, as well as for creating decision support systems in this field. Hence, the research presented in this paper holds substantial practical significance in advancing modeling and analysis methods for biotechnological processes. This, in turn, can play an essential role in the development of various biotechnology areas, including bifidobacteria production for the food industry and the creation of new pharmaceuticals.
Keywords: artificial network, perceptron, neural network training algorithm, algorithm, software, mathematical and computer modeling, biotechnological process, bifidobacteria, biotechnological process model
Visitors: 123

14. Neural network diagnosis of the cardiovascular diseases based on data-driven method [№1 за 2024 год]
Authors: Mosin S.G. (smosin@vpti.vladimir.ru) - Vladimir State University named after Alexander and Nikolay Stoletovs, Ph.D;
Abstract: The paper considers methods for diagnosing cardiovascular diseases by electrocardiogram (ECG) tracing using artificial intelligence methods. It also determines the problems of diagnosing cardiovascular diseases by model-driven methods. The author proposes an approach to diagnosing cardiovascular diseases by a data-driven machine learning method without extracting the characteristic parameters of ECG signals. There is a presented architecture of a neuromorphic ECG signal analyzer based on a one-dimensional convolutional neural network, as well as its design route. Experimental studies were carried out on a set of ECG signals PTB-XL; they confirmed the operability and efficiency of the proposed approach. Both structural and parametric synthesis of a neuromorphic analyzer was performed for a different number of internal layers and initial training parameters. A comparative analysis of the obtained results found that a neural network with two convolutional layers has low training accuracy and high diagnosis errors; a three-layer neural network contributes to the growth of type I error; a four-layer neural network contributes to the growth of type II error. The use of a three-layer convolutional neural network with a smaller pooling window provided the diagnosis of up to 85.66 % of myocardial infarction cases. In conclusion, the author gives the directions for further research to improve the diagnosis accuracy by reducing an input ECG signal dimension, as well as introducing a probabilistic assessment of whether the considered signal belongs to one of the possible states of an ambiguity group.
Keywords: NA design route, neuromorphic analyzer (NA), ECG signals, diagnostics of cardiovascular diseases, NA architecture, computer aided design, data-driven method
Visitors: 92

← Preview | 1 | 2