## No 36 (2020)

**Year:**2020**Articles:**13**URL:**https://ered.pstu.ru/index.php/elinf/issue/view/216**DOI:**https://doi.org/10.15593/.v0i36

Articles

ANALYSIS OF IMPACT OF RENEWABLE ENERGY SOURCES WITH POWER CONVERTERS ON THE PROCESSES IN MODERN ELECTRIC POWER SYSTEMS

#### Abstract

One of the promising areas of modern electric power systems development is the use of renewable energy sources in their structure. Among renewable energy sources, the most widespread are installations that include power converters based on static voltage converters and providing connection of these energy sources to the network. At the same time, such installations have dynamic properties that are strikingly different from those of conventional generation sources and, accordingly, have a completely different impact on the power system in general. The purpose of the research is to carry out a preliminary analysis of the impact of penetration of renewable energy sources with power converters in their topology on the behavior of transients in modern power systems. The method of power systems mathematical modeling was used for research, the basis for the implementation of detailed and reliable modeling was a hybrid hardware and software modeling - Hybrid Real-Time Power System Simulator. Results: several options for the penetration of renewable energy sources were considered on the example of a test model of a large-scale power system and the impact of their integration on the rate of change of frequency, the level of post-emergency oscillations in the power system, and the frequency of power swings was determined. The practical significance of the performed studies is that it can be used to develop specific recommendations for power system automation’ tuning, the optimal and efficient operation of which directly depends on the comprehensiveness and reliability of information about a single continuous spectrum of normal and abnormal steady-state and transient processes in modern electric power systems with renewable energy sources.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):7-30

DETERMINING THE SAMPLE LENGTH OF THE ELECTRONIC LETTERS MODEL

#### Abstract

In the conditions of intensive development of various spheres of activity of the state and society, the use of advanced information technologies becomes one of the most important and often decisive factors that determine the effectiveness of all levels of management. Information technologies hold a strong position in everyday life and in various spheres of government activity. Information systems of various types and purposes are being introduced for processing and analyzing information, forecasting and supporting management decisions making at various levels. The functioning of almost any organization in a varying degree depends on how efficiently and steadily its information system works, how reliably its information resources are protected from various information security threats, one of which is spam. It is worth noting that there have been many attempts to solve the problem of its detection once and for all. Research is ongoing in this subject area constantly. Based on its results, various approaches are proposed and implemented in practice. They are mostly based on the well-proven basic statistical processing of various parameters of emails. At the same time, the content of e-mails is not fully taken into account, which leads to an increase in the number of Type I errors. To address this issue, the authors previously proposed a model of emails that takes into account the content of emails, which often changes depending on the tasks performed by users and their changing information needs. Its key feature is that it operates with data obtained from the original text of emails and converted into a numeric vector. As the parameters of the model of emails that influence the selection of emails test segments, which are a reflection of their distinctive features, the authors identified: the number of numeric codes associated with text characters in the function of converting letters to numeric vector; the step of fetching text characters in the function of converting letters to a numeric vector; sample length (length of the «generator» - the sequence that generates the «gene»). Purpose of the research: to assess the influence of the sample length (length of the «generator» - the sequence that generates the «gene»), which is the email model key parameter, on results of the use of the model in spam detection. Methods: use of decision rules of the implicative type «IF ..., THEN ...». Results: the results of the experiment confirm the correctness and applicability of the email model previously developed by the authors. At the same time, the application of the e-mail model gives the best detection results when the numerical value of the sample length (the length of the «generator» - the sequence generating the «gene») is equal to 1 or 2, which is confirmed by the values of the aggregated assessment measure (the value combining the measures of completeness and accuracy of detection and representing their harmonic mean). When the length of the «generator» is greater than or equal to 3, the detection results deteriorate on average by more than 10 %. The inexpediency of using the «generator» length values greater than or equal to 3 is justified by an unacceptable decrease in the values of the detection completeness and the aggregated assessment measure. Practical relevance: the authors' model of emails and an approach to spam detection allow to highlight the sequences of text characters corresponding to the characteristics of a certain class of emails, and to make the spam detection process individual for the recipient.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):31-47

NUMERICAL STUDY OF THERMAL PROCESSES IN A WELL WITH A BOTTOM HEATER

#### Abstract

Due to the decline in natural oil reserves in oil production, the problem of developing fields with high-viscosity oil is very relevant. One of the ways to produce high-viscosity oil is to warm up the bottomhole zone of the oil well using a heater. Purpose: development and study of a model of the heat and mass transfer process in an oil well with a bottom-hole heater by comparing various models of turbulent and laminar flow and identifying the optimal one. Methods: The paper considers a two-dimensional axisymmetric mathematical model of the heat and mass transfer process in an oil well with a bottom-hole heater, which is used for wells with high-viscosity oil. The use of a local heater in the bottom-hole area allows to reduce the viscosity of the oil entering the pump, thereby ensuring the smooth operation of the well. A longitudinal section of an oil well with a depth of 105 meters, surrounded by soil, was considered. The heater is located above the bottom. The numerical implementation of the developed mathematical model was carried out by the finite element method in ANSYS. The ICEM CFD preprocessor was used to build the geometry and mesh the finite elements. The problem was solved by a numerical finite element method in the Fluent engineering environment. Results: a mathematical model of an oil well with turbulent and laminar oil flow was developed, taking into account the bottom-hole heater. As a result of the study, the fields of velocities and temperatures were obtained in the entire volume of the oil well. The distributions of velocities and temperatures over the cross-section of the tubing were constructed and a comparison between various turbulent flow models and the k-epsilon model was made. The influence of the viscosity value on the temperature before entering the electric centrifugal pump was investigated. The most optimal flow model was selected under the given conditions. Practical relevance: the results obtained make it possible to choose the optimal flow model for calculating this problem and to take the necessary measures to reduce the oil viscosity in a previously known area, which will significantly reduce material costs during field development.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):48-62

MODELING THE MOTION OF CONDUCTING PARTICLES IN THE ACTIVE ZONE OF A LINEAR INDUCTOR OF EDDY-CURRENT SEPARATOR

#### Abstract

The improvement of technologies for the collection and primary processing of secondary non-ferrous metals is a prerequisite for the development of secondary non-ferrous metallurgy and the becoming, of solid waste processing enterprises. One such technology is eddy-current (electrodynamic) separation in a traveling magnetic field. Such separators easily fit into technological lines and are used in the extraction of non-ferrous metals from various types of metal-containing solid waste (automobile scrap, mixed production and municipal waste, cable and electronic scrap, etc.), as well as in the processing of complex non-ferrous scrap in preparation it to metallurgical processes. The article discusses eddy-current separators based on linear inductors. The Installation consists of a conveyor that moves the materials to be separated and an inductor located under the conveyor belt. Modeling the interaction between a magnetic field and conductive particles is complicated by the existence of various electromagnetic and mechanical (gravitational and friction) forces. Purpose: Development of a mathematical model taking into account the action of various forces on particles and an assessment of ways to increase the efficiency of eddy-current separation. Results: A mathematical model for studying an eddy-current separator based on a linear induction machine is presented. Separation is the result of combined actions of different forces on conductive particles. Consequently, an important part of the separator model is the equations for the combined electromagnetic and mechanical action. The model takes into account the movement of conductive particles in two directions: the transfer of particles along the conveyor belt and lateral movement under the influence of electromagnetic forces. The results of the study of a prototype of an eddy-current separator based on linear inductors, constructed in the laboratory of the Department of Electrical Engineering and Electrotechnological Systems of the Ural Federal University, are described. The possibility of improving the performance of the separator due to the use of linear induction machines with opposite direction travelling magnetic fields is shown. Calculation results are consistent with industrial test data.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):63-79

FRAUD DETECTION BY MACHINE LEARNING METHODS IN THE CASE OF AN IMBALANCED DATASET

#### Abstract

There are various types of fraud: related to the theft of funds from bank accounts and cards, money laundering, mortgage fraud, etc. One of the fraud types is registering fictitious user accounts and receiving financial rewards from people or companies, those interested in increasing popularity and profitability of an electronic resource. If an application for registration of a presumably fictitious account is found, the electronic resource may subject the application to additional verification. The dataset in the identifying electronic fraud task is usually highly imbalanced: the number of applications for fictitious accounts registration is significantly less than the number of applications for valid accounts registration. Purpose: Creating a classifier using machine learning methods to detect registration of fictitious accounts in the case of using a highly imbalanced training dataset. Results: We considered the existing solutions to the task of combating fraud, witch using machine learning methods. The complex of programs has been written in the Python programming language to solve the task. Data preprocessing and analysis of the features obtained from them were carried out. The paper presents the results of the features correlation analysis and the results of applying various classification algorithms (random forest, bagging, balanced bagging, and LinearSVC) in the task of binary classification of accounts into real and fictitious. Also the results of using various algorithms to reduce the impact on the learning process of the dataset classes imbalance: SMOTE, ADASYN, NCR, Tomek Links, CNN are presented in the paper. The obtained confusion matrices and metrics for the specified algorithms are presented. A comparison of the metrics was made and the algorithm that provides the best classification results was selected. Practical relevance: The use of machine learning methods used in the work can automate the process of detecting fictitious accounts during registration.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):80-95

DEVELOPMENT OF AN ADAPTIVE FUZZY CONTROL SYSTEM FOR RECTIFICATION OF THE RETURN SOLVENT

#### Abstract

Modern technological process is impossible to imagine without automatic control systems. With proper use of such systems, it is possible to obtain optimal operating parameters, both for a particular node and for the technological process as a whole. In the conditions of modern market relations and fierce competition, the problem of high quality of products with minimal production costs becomes especially urgent. This makes it necessary to create high-precision, fast and reliable control of processes and devices. Purpose: development of an adaptive fuzzy control system for the rectification process of a recycled solvent (isoprene-isopentane-toluene fraction) using the Fuzzy Logic Toolbox expansion package for various critical operating situations. The main critical parameters included are the following: high or low temperature, high or low pressure, and high or low hydrocarbon level in the column bottom. Methods: to solve the problem of choosing a critical situation, a fuzzy controller for condition assessmentwas developed. This controller has two resulting outputs regarding the possible critical situations. The first output requires the need to use hot water as a heating agent. The second output requires the need to use steam as a heating agent. Results: depending on the output of the fuzzy controller for condition assessment, one of the two subsequent controllers is switched on. These were in advance set to certain values of input variables. Practical relevance: the proposed method for controlling the process of continuous rectification of isoprene-isopentane-toluene fraction in a tray-type column will allowto timely detect the critical situations during system operation and take necessary measures to remove them, which ultimately will improve the product quality, increase plant capacity and save energy.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):96-116

OPTIMIZATION OF NEURAL NETWORK COMPUTATION

#### Abstract

Microcomputers have computational limitations (processor performance and amount of memory). Running a neural network on such a computer has a lower performance compared to personal and desktop computers. To eliminate this drawback, various methods for optimizing computations of neural networks, in particular, quantization are used. Quantization reduces the precision of numbers that store neural network weights to 8-bit integers or 16-bit floating point numbers with little or no reduction in neural network accuracy, but at the same time reducing the amount of memory used. Quantization makes it possible to optimize computations for a certain type of processor (CPU, GPU and TPU), but this article will only consider optimization for the CPU. Purpose: reducing the file size of the neural network weights. Results: 1) the formulated optimization problem is finding the function of transforming the weights, that reaches the minimum sum of the relative changes in the file size of the weight, the latency, and the accuracy of the neural network, 2) conducted experiments with various types of quantization (dynamic range quantization, quantization using a representative dataset, integer quantization using a representative dataset and float16 quantization), 3) showing computational units and changes in the neural network for the considered types of quantization, 4) a comparative table for these quantizations with various characteristics (the accuracy of the neural network after applying quantization, the neural network latency and the size of neural network weight file), 5) the chosen quantization method reaching the minimum of the optimization function. Practical relevance: reducing the time for updating the neural network on a remote microcomputer. As a result, we chose float16 quantization, since it halves the amount of memory used, and the accuracy of the neural network and performance remains unchanged.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):117-130

THERMAL FIELD MODELING OF THE ELECTRON BEAM WELDING PROCESS WITH DYNAMICALLY VARIABLE PARAMETERS OF PRODUCT’S ALLOYS

#### Abstract

The article deals with the problem of modeling the thermal zone in the process of electron-beam welding, which is widely used at machine-building enterprises of the Russian Federation. The relevance of this study is due to the complexity of the selection of an effective set of technological parameters for titanium alloy products. The use of mathematical modeling methods can significantly reduce time, labor, and economic costs for improving and also working out existing technologies. At the same time, the use of dynamically changing thermophysical parameters in the process of modeling helps to increase the accuracy of modeling. Purpose of the study: improving the efficiency and quality of mathematical modeling of the temperature field. Methods: the approach proposed in the article for the use of dynamically changing thermophysical parameters uses the methods of regression analysis, which make it possible to restore the values of such parameters for alloys of products in the process of modeling. The following widely used approximation methods were used: linear, quadratic, cubic, exponential, logarithmic, which allow restoring the dependence of the missing values of thermophysical parameters depending on the temperature of the material. To simulate the distribution of the temperature field, the apparatus of the theory of welding processes is used, on the basis of which the computational relations are realized. Results: In the process of experimental studies of the application of the approach proposed in the work, the simulation of energy distribution during electron-beam welding of a product element made of VT-14 titanium alloy was carried out. The use of dynamically changing thermophysical parameters of the product made it possible to obtain the energy distribution function in the heating zone in a more natural form, when heating of colder zones does not give a high distribution of heat to adjacent areas. Discussion: the results obtained can be used to improve the quality of the welded joint, as well as to ensure the repeatability of the technological process of electron beam welding.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):131-145

MODELING THE HEAT TRANSFER PROCESS DURING SILANE CROSSLINKING IN A TWO-COMPONENT WATER-GAS MEDIUM USING A VOLUME OF FLUID MODEL

#### Abstract

Polyethylene (PE), a thermoplastic polymer, has the ability to melt and form at elevated temperatures. For this reason, standard grades of polyethylene are limited in their application by the limits of such maximum temperatures, which will be significantly lower than the melting point. Due to crosslinking, the thermal dimensional stability of polyethylene (PEX) can be significantly increased. This makes it possible to use PEX products in applications with higher temperature requirements that exceed the temperature limits for normal thermoplastic grades. Purpose: creating a mathematical model of the heat and mass transfer process in a bath for silane crosslinking of cable insulation filled with hot water with a continuous supply of superheated steam, determining the heating time of the insulation. Methods: Numerical implementation of the developed mathematical model was carried out by the finite element method in the ANSYS, Fluent environment. The VOF model is used to describe the position of the vapor - air interface; the classical ke model is used to describe the turbulent flow. The ICEM CFD preprocessor and quadratic elements were used to construct the geometry and divide it into a grid of finite elements. Results: a non - stationary two-dimensional mathematical model of heat and mass transfer processes in a bath with a cable and a two-component medium-steam-water is developed. The efficiency of the proposed model is analyzed. As a result of the study, a vector field of velocities, temperature fields, and phase distributions in a silanol bath were obtained. The dependence of the temperature of the cable, which is a monolith with averaged properties, on the heating time in the center of the monolithic insulation is constructed. The adequacy of the mathematical model is evaluated and the possibility of its use in practice is determined. Practical significance: The results of this work can be used when creating technological modes of crosslinking for new insulation, when changing the design features of the workpiece and its length on the drum. Using a digital model allows you to estimate the time of the silane crosslinking process for medium-voltage cables and determine the rational modes of the technological process.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):146-158

ONE OF THE APPROACHES TO COMPUTER ATTACKS’ RECOGNITION WHEN IMPLEMENTING THE STATE’S INFORMATION FUNCTION ON THE INTERNAL LEVEL

#### Abstract

The growing impact of the information sphere on social life challenges the State with a new direction of activity - the implementation of the information function. To implement this function at the internal level, public authorities create state information systems that are used for processing sensitive data. Therefore ensuring information security when processing data in these systems becomes crucial for the state. Purpose: to counter targeted attacks on cloud computing based information systems on the example of private cloud environment. Methods: this study offers an approach to recognizing attacks on state information systems based on functional analysis methods, pattern recognition theory, and the use of risk assessment methods to reduce the number of false positives. Results: aт approach for attacks’ recognition for private cloud environment based on the pattern recognition theory is proposed. The attack is presented as a pattern that must be recognized during the processes of storing, processing, and transferring information in the guest operating system. To detect an attack, it is necessary to determine the probability of matching a certain class of attacks of a set of attacks to each object from the set of attacks based on the available description and information about the attacks’ classes. Then functions to estimate the proximity (for example, Chebyshev distance) of the detected attack object to a known attack class can be applied. In addition to the proposed method, an approach based on the information risk assessment is proposed to reduce the number of false positives. The essence of the approach is that the signs of attacks are ranked by the level of influence on certain components of an information system. Each of the ranking criteria is ranked with a certain number of points, which assesses whether a particular criterion is related to the fact of the attack implementation. If the sum of points exceeds a certain threshold, then a fact of attack recognition can be registered. Practical relevance: the results can be used to create intrusion detection systems for information and telecommunications systems, including those used when implementing the State’s information function.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):159-174

SOLUTION OF THE PROBLEM OF OPTIMAL POWER DISTRIBUTION OF INDIVIDUAL COMPENSATING DEVICES FOR A GROUP OF ASYNCHRONOUS MOTORS FOR A STABLE OPERATING MODE AND FOR A GRINDLY VARIABLE LOAD

#### Abstract

Petrochemical, oil refineries, tire factories are characterized by large production areas, they have mainly radial power supply circuits, through which a huge number of 0.4 kV asynchronous motors with a power of up to 200 kW are supplied. They are powered with copper or aluminum cables, which, given their length, have significant resistance. In this regard, the losses of active power during the transmission of electricity to asynchronous motors acquire tangible values, and it becomes necessary to solve the problem of reducing them. The purpose of the study: reduction of active power losses in 0.4 kV cable lines at petrochemical and oil refineries. Methods: it is proposed to solve the optimization problem according to the criterion of minimum active power losses in the radial power supply scheme by means of optimal distribution of reactive power of a given value between compensating devices. A single-line diagram of the power supply of a group of pumps of a technological unit of an oil refinery is considered, a mathematical model of the optimization problem is compiled according to the criterion of the minimum active losses generated during the flow of reactive power in 0.4 kV cable lines. The mathematical model includes an objective function, boundary conditions and constraints. The analysis of possible ways of solving a system of linear algebraic equations by the number of mathematical operations for finding the optimal values of reactive power values is presented. Results: the optimization problem was solved according to the criterion of the minimum active losses in the cable lines of the radial power supply of the group of pumps of the technological unit, the optimal values of the reactive power of the capacitor units connected to asynchronous motors and providing the given power factor on the bus sections were obtained. Losses of active power in cable lines supplying asynchronous motors are calculated with optimal power distribution of capacitor units for various modes of motor load. Practical significance: solving the problem of optimal distribution of reactive power compensation devices leads to a decrease in active losses in the radial power supply circuit of the pumps of the technological unit, as well as in power transformers and higher elements of the power supply circuit, which ultimately leads to a decrease in the cost of manufactured products. A schematic diagram of an automated electrical complex for a group of asynchronous electric motors with an optimal power distribution of compensating devices for a group of pumps of a technological unit has been developed.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):175-192

DEVELOPMENT OF METHOD FOR ASSESSING THE ECONOMIC EFFICIENCY OF THE INFORMATION SECUTITY SYSTEM FOR COMMERCIAL ENTERPRISES

#### Abstract

Modern systems are vulnerable to threats because of insufficient assessment of the information security effectiveness. This problem characterizes a breach of the information security management process. It is obvious that the lack of a scientific base of methods and assessment algorithms leads to a violation of protection effectiveness. From the standpoint of business impact, the cost-benefit assessment identifies vulnerable indicators, ensuring the reliability of business processes. Therefore, the scientific task is to ensure the process of information security system functioning by assessing economic efficiency. Purpose: development of an up-to-date method for assessing the economic efficiency of an information security system based on model of economically efficient functioning of the system. Methods: analytical and mathematical models of the linguistic description of an effective protection system determine the assessment of economic efficiency. The method based on it consists 9 consecutive stages of assessment. It implements an organizational and mathematical algorithm based on the hierarchy analysis method. Results: the article presents the results of designing a linguistic model in the form of an undirected graph. It selects subsystems, requirements and information security tools to ensure the maximal economic efficiency of the system. Graph carries out a preliminary calculation of the assessing economic efficiency results. Article provides the development of the assessment method connected with economical model. The developed method tested on a real protected object. Economical method calculated indicators of the information security system efficiency. As a result, method calculates economic efficiency data, the degree of risk for information system, as well as model parameters. It gives a verbal description of the effective security system parameters including the protection coefficient. Practical relevance: the developed method tested and applied in the framework of improving the performance of the security system at the facility based on commercial enterprise in Perm Krai. The article also provides recommendations of using the method in organizational measures for information security. Moreover, the authors offered the ways of implementation an economic efficiency assessment in the form of a multi-criteria approach.

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):193-213

THE USE OF DIRECTED GRAPHS IN THE MANAGEMENT PROCESSES OF ECO-ECONOMIC ACTIVITIES

#### Abstract

Management of eco-economic activities under modern conditions plays an important role in ensuring the effective functioning of both individual industrial enterprises and the region as a whole. This situation requires a search not only for ways to improve the organizational and economic mechanism of environmental management, but also its individual tools, adapted to the features of a specific management object. Directed graphs can be considered as such a tool, which allow creating simulation models of various environmental and eco-economic processes and systems. The vertices of an orgraph can be used to store the necessary information about the objects of study, for example, about the local rate of environmental or eco-economic security. Purpose of the study: to analyze efficient ways of using directed graphs in the management of eco-economic activities and to adapt existing developments to the processes of ensuring eco-economic safety of industrial enterprises and a region. Results: the analysis of known organizational and technological solutions in the field of control of systems of different levels using directed graphs was carried out. Graphs of possible strategies for achieving the required environmental safety rate in the region have been developed, taking into account the interests of enterprises-users of natural resources and ensuring sustainable socio-ecological and economic development of the territorial entity. The optimal trajectory of movement has been determined, which ensures the required eco-economic safety rate, taking into account the minimum costs for its achievement. The difference between the approach used lies in the construction of a directed graph, taking into account the features of the investigated ecological and economic system, as well as the type of work required to achieve a given level of environmental and economic security and their cost estimate. Practical significance: the proposed mechanism for the considered eco-economic systems (region - industrial enterprises) management is recommended for its application by the management structures of the region and the management of enterprises that have a significant negative impact on the environment. It is advisable to use the obtained results when developing strategies for sustainable development of territorial entities, enterprises and their associations (clusters, holdings, financial and industrial groups and other integration forms).

**PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems**. 2020;(36):214-231