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

Bookmark

Next issue

1
Publication date:
16 March 2020
-->

Journal articles №4 2019

11. A method of identifying technical condition radio engineering means using artificial neural network technologies [№4 за 2019 год]
Authors: Dopira R.V. (rvdopira@yandex.ru) - NPO RusBITex (Professor, Head of Department), Ph.D; A.A. Shvedun (ashvedun@mail.ru) - Military Academy of the Aerospace Defense (Postgraduate Student); D.V. Yagolnikov (yagolnikov_dv@mail.ru) - Military Academy of the Aerospace Defence (Lecturer); I.E. Yanochkin (i.yanochkin@rusbitech.ru) - Russian Basic Information Technologies (Head of Department), Ph.D;
Abstract: Due to the fact that modern military-grade radio equipment is becoming functionally and technologi-cally more complicated, the urgency of the task of creating functional control systems and identifying technical state of radio equipment is increasing. Nowadays, there are no effective and fully automatic systems for identifying technical state of vari-ous types of radio equipment. One of the ways to solve the problem is to create systems for identifying technical state of radio equipment is based on machine learning principles. A distinctive feature of the application of trained artificial neural networks to solve the identifying problem is the development of a prototype of the observed situations, generalizations for the predomi-nance and similarity in a variety of same type radio equipment, as well as high efficiency and reliabil-ity of solving this problem. The paper presents a method for identifying technical state of radio equipment using case-law prin-ciples of machine learning of artificial neural networks. It allows solving the problem of identifying current classes of the radio equipment technical condition based on measurement results of the main system controlled parameters in real time. Taking into account the problem specifics, the choice of a multilayer direct distribution neural net-work including three hidden layers is substantiated. The number of neurons of the input layer is deter-mined by the number of controlled parameters of the technical condition of the main systems of radio equipment of a particular type. The number of output layer neurons is determined by the number of possible classes of the radio equipment technical condition. Elementary converters of this network have an activation function of a sigmoid type. To train an artificial neural network, the authors used a heuristic modification of the Levenberg-Marquardt algorithm.
Keywords: identification of a technical condition, artificial neural networks, radio engineering tools, machine learning, data normalization, heuristic modification of the levenberg-marquardt algorithm
Visitors: 850

12. The development of fast software implementation of specialized neural network architecture with sparse connections [№4 за 2019 год]
Author: Yu.S. Fedorenko (fedyura11235@mail.ru) - Bauman Moscow State Technical University (Postgraduate Student);
Abstract: The paper is devoted to the development of fast software implementation of a specialized neural net-work architecture. Feature engineering is one of the most important stages in solving machine learning tasks. Nowadays, the algorithms of handcrafted feature selection lose their popularity, giving way to deep neural networks. However, application of deep models is limited in online learning tasks as they are not able to learn in real time. Besides, their using is difficult in high-loaded systems due to signifi-cant computational complexity. In the one of previous articles, the author has proposed a neural network architecture with automat-ic feature selection and the ability to train in real time. However, specific sparsity of connections in this architecture complicates its implementation on the base of classic deep learning frameworks. Therefore, we decided to do our own implementation of the proposed architecture. This paper considers data structures and algorithms developed when writing software implementa-tion. It describes sample processing in details from the program system point of view during model pre-dicting and training. For a more complete description of implementation details, there are UML classes, sequences and activity diagrams. The performance of the developed implementation is compared with implementations of same ar-chitecture on the base of deep learning frameworks. The analysis has shown that the developed soft-ware works an order of magnitude faster than library-based implementations. Such acceleration is due to the fact that the developed implementation is optimized for a specific architecture, while the frame-works are designed to work with a wide class of neural networks. In addition, the benchmarks have shown that the developed implementation of a proposed neural network works only 20-30 percent slower than a simple logistic regression model with handcrafted features. Thus, it can be used in high loaded systems.
Keywords: benchmark, index reuse, modular arithmetic, sparse connections, categorical features, deep neural networks, feature engineering
Visitors: 765

13. Features of using neural network models to classify short text messages [№4 за 2019 год]
Authors: Dli M.I. (midli@mail.ru) - (Smolensk Branch of the Moscow Power Engineering Institute, Ph.D; O.V. Bulygina (baguzova_ov@mail.ru ) - Smolensk Branch of the Moscow Power Engineering Institute (Associate Professor), Ph.D;
Abstract: Nowadays, public authorities are actively developing technologies of electronic interaction with organ-izations and citizens. One of the key tasks in this area is classification of incoming messages for their operational processing. However, the features of such messages (small size, lack of a clear structure, etc.) do not allow using traditional approaches to the analysis of textual information. To solve this problem, it is proposed to use neural network models (artificial neural networks and neuro-fuzzy classifier), which allow finding hidden patterns in documents written in a natural lan-guage. The choice of a specific method is determined by the approach to forming thematic headings: convolutional neural networks (for unambiguous definition of rubrics); recurrent neural networks (for significant word order in the title of rubrics); neuro-fuzzy classifier (for intersecting thesauri of ru-brics).
Keywords: text categorization, neuro-fuzzy classifier, artificial neural networks
Visitors: 801

14. Peculiarities of porting the Robot Operating System framework onto Elbrus platform [№4 за 2019 год]
Authors: A.A. Tachkov (tachkov@bmstu.ru ) - Research and Training Centre "Robotics" Bauman Moscow State Technical University (Head of the Department "Automated transport systems"), Ph.D; A.Yu. Vukolov (twdragon@bmstu.ru) - Research and Training Center "Robotics" Bauman Moscow State Technical University (Programmer); A.V. Kozov (alexey.kozov@gmail.com) - Research and Training Centre "Robotics" Bauman Moscow State Technical University (Engineer);
Abstract: The Robot Operating System is the most widespread helper framework for mobile robots control sys-tems development. However, it fully supports only Ubuntu/Debian Linux-based software, which leads to limitation of the possibility to use calculating equipment developed natively in Russia within design control systems. The authors ported ROS Melodic Morenia onto Elbrus platform natively developed in Russia (computer based on Elbrus-4C CPU). This paper describes main peculiarities of the porting process associated with the differences be-tween Elbrus operating system and most of Linux distributions. Version matching for Elbrus integrated software on which the ROS depends is also considered. The authors attempted to determine ROS build dependency libraries that could be fully installed on Elbrus using only system packages and to repack these libraries into deb-packages for further installation onto similar computers. In addition, the paper describes the developed deployment scenarios for ROS ready-out-of-the-box. There is a description of testing of the prepared distribution performed for the task of multilayered passability map building. This task was solved on control system with LIDAR point cloud input. The authors demonstrate the results of tests as treatment times for the identical point clouds processing and map layer refreshment. The tests were run on systems based on Elbrus-4C, Intel Core i3-3220 and Intel Core i7-6700HQ CPUs using the same ROS version. The authors conclude that the ROS Melodic More-nia deployed onto Elbrus platform iы fully operational.
Keywords: mobile robot, point cloud, layered map, performance testing, building software packages, porting, ros, elbrus platform
Visitors: 890

15. Earth's surface visualization in simulation systems [№4 за 2019 год]
Authors: Giatsintov A.M. (algts@inbox.ru) - SRISA RAS; Mamrosenko K.A. (kirillam@ya.ru) - SRISA RAS, Ph.D; P.S. Bazhenov (bps@niisi.ras.ru) - Center of Visualization and Satellite Information Technologies SRISA (Programmer);
Abstract: Planet visualization is used in many fields: development of geographic information systems, multime-dia systems, simulation systems and simulators. The present paper describes approaches to displaying the Earth's surface that provide real-time visualization. It also lists the problems that arise during ex-tended landscapes visualization and are related to productivity, the coordinates system transformation and visualization accuracy. The paper presents an approach to the Earth's surface visualization that is based upon the use of clipmaps. It allows simplifying data preparation for various parts of the Earth's surface, as well as re-ducing the number of prepared data sets for both the underlying surface and textures with height data. To implement the proposed approach, the architecture of the component used to generate and visu-alize the Earth's surface was developed. The component is built into existing visualization systems. This architecture has the following advantages. It is not necessary to develop own specialized visuali-zation system from scratch. It is possible to use existing visualization systems both open and proprie-tary. Computing load is distributed across available threads in the thread pool. Creation of drawing calls that will be sent to the visualization system is one of multi-threaded pro-cessing examples. The algorithm for processing incoming drawing calls depends on implementation of integration with the particular visualization system. Component's architecture provides the ability to set the time during which the data will be processed inside the component. After each operation is fin-ished the time taken to complete the operation is calculated and the ability to continue processing other tasks is determined. If the time limit is exceeded, task processing will be terminated until the next com-ponent update call.
Keywords: coordinate system, simulation systems, render, terrain, wgs84
Visitors: 795

16. Method and software for intellectual support of making logistics decisions [№4 за 2019 год]
Authors: Borisov V.V. (BYG@yandex.ru) - Smolensk Branch of the Moscow Power Engineering Institute, Ph.D; A.V. Ryazanov (ballack.139@mail.ru ) - Smolensk Branch of the Moscow Power Engineering Institute (Postgraduate Student);
Abstract: The paper proposes intellectual support method for making logistic decisions allowing to solve the complex of the following problems: to determine required resources in the course of orders distribution by territorial zones; to divide logistics service territory into zones based on genetic clustering; to dis-tribute orders by zones according to their purpose; to fuzzily assess and assign logistics facilities to fulfill orders based on the modified G. Kuhn method. To implement the stages of intellectual support method for making logistic decisions a set of proce-dures has been developed, and namely procedures: to divide the service territory into zones based on genetic clustering; to determine the required number of logistics facilities (resources) when distributing orders by zones based on a moving time window; to distribute orders by zones and to assign logistic fa-cilities to fulfill orders based on the modified G. Kuhn method. Distribution of orders by zones evaluation and logistic facilities assignment are based on the inte-gral indicator of logistic facilities and orders degree of compliance, considered in the paper. To deter-mine the indicator, a cascading fuzzy production model has been developed that allows taking into consideration different-quality characteristics of logistics facilities and various strategies for orders distribution. Algorithms and software have been developed implementing the proposed method of logistic deci-sions intellectual support. The software includes: (1) subsystem for division of territory into zones, consisting of genetic clusterization and zonal division program modules; (2) subsystem for orders dis-tribution and logistics facilities assignment, consisting of program modules to determine the required number of logistics facilities for zones, to assess conformity between logistics facilities and orders, and to assign logistics facilities. A comparative assessment was carried out, confirming the improved quality of logistics decisions made using the proposed method and software.
Keywords: fuzzy production model, genetic clusterization, logistics facility, orders distribution, division of territory into zones, intellectual support for making logistics decisions
Visitors: 766

17. Model and algorithm of selecting software architecture for IoT systems [№4 за 2019 год]
Author: Yu.V. Yadgarova (y.v.yadgarova@gmail.com) - Bauman Moscow State Technical University (Postgraduate Student);
Abstract: The paper presents an analytical model for cost estimation and an algorithm for selecting a basic pat-tern of software architecture, as well as design tactics for the IoT (Internet of things) systems. It gener-alizes the concept of IoT technologies, reviews software systems quality parameters, emphasizes the main significant parameters applicable to IoT systems and proposes methods to achieve them. The nec-essary quality parameters of software systems are achieved by implementing the basic pattern of soft-ware architecture and related design tactics. The paper introduces the analytical model of dependency of project labor intensity on the used el-ements of software architecture. The project labor intensity is calculated according to the COCOMO II method. The search algorithm for the basic architecture pattern and design tactics is presented. The in-dicated algorithm is built on the basis of a local search when solving the problem of satisfying con-straints with minimizing the labor intensity function. Selecting a pattern, user preferences are also tak-en into account. The model and algorithm allow selecting the most appropriate architecture patterns and tactics for a particular type of project at early stages of their design. Indicated approach allows reducing errors in software architecture design at the initial stage of IoT-architecture pattern selection. The paper considers implementation of this approach in the project of flexible workspace management system development. It is advisable to use the approach to achieve the required system quality parameters and minimize errors when selecting software architecture at the ini-tial stages of the project, finally reducing the project cost. The approach can also be used to develop fully functional prototypes on a tight schedule.
Keywords: csp-task, architecture pattern, design tactics, quality parameters of software architecture, software architecture design, iot architecture, quality of IoT systems
Visitors: 744

18. Component in the Kotlin programming language for integrating executable programs into internet resources [№4 за 2019 год]
Authors: A.V. Prendota (prendota@mail.ru) - IntelliJ Labs, The National Research University of Information Technologies, Mechanics and Optics (Student); P.V. Balakshin (pvbalakshin@gmail.com) - The National Research University of Information Technologies, Mechanics and Optics (Associate Professor), Ph.D;
Abstract: The paper provides an overview of tasks, methods and tools for training and working with the Kotlin programming language. This language was created as an alternative to other popular programming lan-guages based on JVM (Java Virtual Machine), and also to the languages that were used to write various Android and iOS applications. Due to the fact that Kotlin leveled inconveniences of competitive pro-gramming languages, it has become the official tool for the Android operating system since 2017. The importance of creating and supporting an online programming environment as a full-fledged training resource is shown. Using the online environment in training courses makes it possible to create sample programs for studying stylistic features of the language in the form of code that is executed di-rectly in the browser. Such approach allows ensuring the development of new projects even in the ab-sence of a full-fledged development environment on the developer's computer. Syntax highlighting and code completion also attract new users. The paper presents a brief analysis of existing online development platforms and highlights their shortcomings associated with integration problems with third-party sites, a small number of illustrative examples, and the lack of syntax highlighting and code completion. Furthermore the paper provides in-formation on how to solve a number of issues associated with writing and executing code in the Kotlin language. The paper considers the use of the Kotlin Playground library, which converts HTML blocks into specific code editors. This makes it possible to execute the created editors directly in the browser. Kotlin Playground library features such as code execution and compilation for various platforms, markup, code completion and highlighting capabilities, test script creation and execution, and working with the API are also considered. These features allow users to adapt every component of the online environment to his/her needs and to correctly integrate the resulting environment into Internet re-sources. The paper gives examples of the Kotlin Playground library functions, presents its syntax, installa-tion and further loading rules using Node Package Manager, as well as the use of this library. The paper concludes with a link to documentation and identifies common Russian and international educational online platforms on which Kotlin training courses have been or are being implemented using the Kotlin Playground library.
Keywords: syntax highlighting, code completion, online development environment, api, online platform, jvm, browser, kotlin
Visitors: 761

19. Implementing an expert system to evaluate technical solutions innovativeness [№4 за 2019 год]
Authors: Ivanov V.K. (mtivk@mail.ru) - Tver State Technical University, Ph.D; Obraztsov I.V. (sunspire@list.ru) - Tver State Technical University, Ph.D; Palyukh B.V. (pboris@tstu.tver.ru) - Tver State Technical University, Ph.D;
Abstract: The paper presents a possible solution to the problem of algorithmization for quantifying innovativeness indicators of technical products, inventions and technologies. The concepts of technological novelty, relevance and imple-mentability as components of product innovation criterion are introduced. Authors propose a model and algorithm to calculate every of these indicators of innovativeness under conditions of incompleteness and inaccuracy, and sometimes inconsistency of the initial information. The paper describes the developed specialized software that is a promising methodological tool for using in-terval estimations in accordance with the theory of evidence. These estimations are used in the analysis of com-plex multicomponent systems, aggregations of large volumes of fuzzy and incomplete data of various structures. Composition and structure of a multi-agent expert system are presented. The purpose of such system is to process groups of measurement results and to estimate indicators values of objects innovativeness. The paper defines ac-tive elements of the system, their functionality, roles, interaction order, input and output interfaces, as well as the general software functioning algorithm. It describes implementation of software modules and gives an example of solving a specific problem to determine the level of technical products innovation. The developed approach, models, methodology and software can be used to implement the storage technology to store the characteristics of objects with significant innovative potential. Formalization of the task's initial data significantly increases the possibility to adapt the proposed methods to various subject areas. There appears an op-portunity to process data of various natures, obtained during experts’ surveys, from a search system or even a measuring device, which helps to increase the practical significance of the presented research.
Keywords: certificate, estimation, invention, implementability, relevance, expert system, data warehouse, term, innovation
Visitors: 804

20. Method of forming a priority list of automated control equipment in special purpose systems and its software implementation [№4 за 2019 год]
Authors: V.L. Lyaskovsky (l_vik_l@mail.ru) - Research Institute of Information Technologies (Professor), Ph.D; I.B. Bresler (niiit@niiit.tver.ru) - Research Institute of Information Technologies (Associate Professor, Director General), Ph.D; M.A. Alasheev (niiit@niiit.tver.ru) - Research Institute of Information Technologies (specialist), Ph.D;
Abstract: The paper considers the method of forming a priority list of control equipment for distributed infor-mation management systems (DIMS) designed for special and military applications that have to be equipped with automation tools, as well as its software implementation as a part of decision making support system. The need to develop and apply this method arises from the fact that DIMS, as a rule, are created in several stages over a long time. This is mainly due to high complexity and cost of development, manu-facture and supply of automation equipment complexes, as well as to limited financial resources, tech-nological and production capabilities of all participants of this process. At the same time it is intuitive-ly clear that equipping some controls with automation tools can make a more significant contribution to improving the efficiency of the entire system than automating other controls. However, there has been no formalized method till present that could substantiate the sequence of equipping controls with au-tomation facilities based on their most significant parameters and characteristics. In this regard devel-opment of a method for forming the priority list of DIMS control equipment is a very important and practically significant task. The essence of the proposed method lies in the consistent assessment of every unit of control equipment (CE) in accordance with the developed system of classification criteria. Moreover, all clas-sification criteria are hierarchically interconnected. Their importance decreases from the first to the last one. Application of the method is connected with the need to collect, store and process arrays of initial data. To use the method in a more convenient way, to reduce the time of information processing and the number of errors associated with the human factor, the authors developed the software implementing the method as an integral part of the developed decision making support system. The method can be used by contractors and research organizations to substantiate the sequence of work in the course of DIMS development and elaboration.
Keywords: equipment with automation tools, automation facilities set, life cycle of automated control systems, control agency, an automated system issue
Visitors: 795

← Preview | 1 | 2 | 3 | Next →