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


Next issue

Publication date:
16 September 2020

Articles of journal № 2 at 2020 year.

Order result by:
Public date | Title | Authors

1. Software complex SkillsForYou for the IT specialists training management [№2 за 2020 год]
Authors: Borisov V.V., S.P. Yanukovich, T.V. Mrochek , D.S. Orekhovsky
Visitors: 971
Currently, there are options for training IT specialists in the educational services market. They over-whelmingly do not take into account the student’s personal characteristics, as well as the market needs in labors. To solve these problems, the authors have developed software complex SkillsForYou (in the Python programming language), which has a modular structure that allows implementing support for the IT specialists training based on the cycle of automated management of training IT specialists (based on the theory of organizational systems management and consisting of stages of planning, organization, stimulation, control). The SkillsForYou contains a large number of training courses of different complexity in various languages and programming technologies (Java, C#, Python, JavaScript, HTML, CSS, PHP). The complex includes a module for analyzing the personal and psychological learner’s qualities, which contains many different psychological tests to assess the quality formation-level necessary for IT specialists, and the specialist team formation to collaborate on projects. The mathematical module on the psychological testing results and the results of calculations using swarm intelligence algorithms determine recommendations for organizing the learning process for each learner. Swarm intelligence algorithms use statistics on the complexity of the problems selected for solving. Problems for training courses are in accordance with the taxonomy of Bloom's training goals, and to assess the problem complexity to solve, the course teacher assigns a cost to the problems. In order to facilitate the search for applicants by HR specialists and employers the software com-plex stores the education history in the learning courses and the learners’ rating.

2. Architecture of predictive maintenance system of complex multi-object systems in Industry 4.0 concept [№2 за 2020 год]
Authors: Sai Van Cuong , M.V. Shcherbakov
Visitors: 1042
Properly formed strategy of maintenance of equipment plays a critical role in modern economic condi-tions characterized by crisis phenomena and high levels of competition. Recently, as part of the implementation of the concept of Industry 4.0 in the field of maintenance of complex multi-object systems, the most promising approaches are based on the use of advanced meth-ods for analyzing large data based on innovative artificial intelligence technologies. It is mainly about the concept of predictive maintenance (PdM), namely the creation of predictive models to prevent equipment failures. This maintenance strategy allows to move from time-based maintenance to condi-tion based maintenance, taking into account the prediction of changes in system states in order to achieve their maximum performance at minimal cost. Therefore, this paper discusses the key elements for implementing the PdM strategy. As a result, an architecture for predictive maintenance of complex multi-object systems in the con-cept of Industry 4.0 is proposed. The proposed system includes three modules: an offline-analysis module for accumulated data, an online-analysis module for streaming data, and a decision support module. The main functions of the first two modules are early detection and prediction of equipment failure based on machine learning methods. Based on the information received from the online analysis module, the decision support module generates optimal decisions when choosing a strategy for influ-encing the equipment, if necessary. Such solutions maintain an optimal balance between the cost of performing technological impacts and the magnitude of potential damages and risks from equipment failure.

3. The specialization problem in hierarchical learning control systems for the foraging problem [№2 за 2020 год]
Authors: E.E. Ovsyannikova , M.A. Rovbo
Visitors: 865
The paper considers the specialization problem in a multi-agent system that consists of identical agents capable of learning that also has weakly connected sub-goals in the problem. The authors propose and consider a hierarchical control system, which is based on ε-greedy Q-learning, with learning on the foraging problem with different types of resources. A group of initially identical agents face the problem of collecting several types of environmental resources and bringing them to a base. The resources must be collected in a certain proportion and the reward is only for transporting the missing resource to the base. The proposed control system has several levels. The upper level is responsible for choosing the re-source type that the agent is going to bring to the base. The lower one selects the learning algorithms and determines the skill of collecting a certain type of resource. Thus, in such multi-agent system, the hierarchy in control allows initially identical agents to form different skills of collecting resources dur-ing their work. The hypothesis under consideration is that, despite the problem does not have prerequisites for op-timization through specialization (ideally working agents could collect different resources alternately), the difference in experience gained during operation can make more efficient specialization of the dif-ferent agents on different resources. The results were obtained using a computational experiment on a computer model. The effect of specialization was observed under certain conditions. Also, specializa-tion and the introduction of a hierarchy in the control system have notably stabilized the performance efficiency of the multi-agent system.

4. A software product development for building a value stream map [№2 за 2020 год]
Authors: S.A. Gunkov, S.S. Akimov
Visitors: 866
The paper discusses the construction problems for a value stream flow map. The authors determined that this card is an element of “lean manufacturing” and allows us to visualize material flows, high-lighting those that add value to the created product. The authors noted that the manual construction of such cards is a complex and routine work, and the relative simplicity of the process mapping makes it possible to algorithmize them, which is the prerequisite for developing a software product that imple-ments value stream maps. The paper provides an existing software product overview. These products have in their function-ality the ability to build value stream flowcharts. There are the main characteristics and functional fea-tures, as well as a system for evaluating the data of software products. The authors determined that de-spite the variety of existing software solutions, currently there is no software product that implements a flow map with an interface in Russian. In addition, each software product is quite expensive, which limits its use. Therefore, The authors decided to develop their own software product on the basis of the Aerospace Institute of Orenburg State University. As a result, there was new software, made in Russian, which is highly specialized and performs all the necessary functions for constructing flow maps.

5. Trapper: an operating system bootstrapping package for IBM PC compatible computer systems [№2 за 2020 год]
Author: Y.I. Klimiankou
Visitors: 819
The paper presents an overview of the bootstrapping process on the IBM PC-compatible computer systems and proposes an architecture of the operating system bootstrapping package. The proposed package implements a framework for constructing boot images targeted at non-traditional operating systems like microkernel, an exokernel, unikernel, and multikernel. The bootstrapping package con-sists of three sets of independent boot modules and a Boot Image Builder application, which creates OS boot images. This application integrates and chains boot modules with one another to organize a com-plete bootloader chain. They are necessary to bring the operating system to a working state. The boot-strapping package architecture reflects the principal stages of the computer system boot process. Each set of boot modules is connected to the particular boot stage and forms a layer that is responsible for performing its own clearly defined set of functions and relies on clearly defined inter-layer interfaces to strictly isolate dependencies on the boot device, firmware and the specifics of the bootloaded oper-ating system. The paper presents the implementation of the described architecture for boot image generation de-signed and implemented for a research multikernel operating system and explains how it boots up. Additionally, the paper proposes the full separation idea of initialization code out of the operating system kernel and its movement into the independent OS loader module. Following this idea leads to the exclusion of the “dead” initialization-related program code from the OS kernel. In the commodity operating systems, such code runs only once during system boot, however, being the part of the kernel executable binary image, continues to occupy memory until the system shuts down.

6. Machine learning to predict the supercomputer jobs execution time [№2 за 2020 год]
Authors: A.V. Baranov, D.S. Nikolaev
Visitors: 812
The authors devoted the paper to machine learning methods and algorithms for the supercomputer job execution time predicting. The supercomputer systems statistics for multiple-access shows that the ac-tual execution time for most jobs substantially diverges from the time requested by the user. This re-duces the scheduling job efficiency because inaccurate estimates of execution time lead to a subopti-mal schedule for job launches. The paper considers the job classification, which is based on actual time for job performance to the requested one. There were six job classes moreover the ratio of the actual job completion time to the requested time for each class differs by an order of magnitude from the previous class. Statistical data from a shared-use supercomputer system is the basis for predicting the job completion time by assign-ing the incoming job to one of the classes. The supercomputer RIKEN Integrated Cluster of Clusters (RICC) statistics presented in SWF format were as initial data. The analyzing results for these statistics allowed us to identify significant features of the job flow for machine learning. There was a feature ranking by importance and there were hidden patterns that affect the forecasting accuracy, in particu-lar, there was a mutual correlation for selected features. For common machine learning algorithms, such as logistic regression, decision tree, k nearest neighbors, linear discriminant analysis, support vector method, random forest, gradient boosting, and direct propagation neural network, there were estimates of the probability of a correct forecast. Algo-rithms of the decision tree, random forest, and direct propagation neural networks showed the best val-ues.

7. Development of a geoinformation systems protecting method and spatial data based on a neural network [№2 за 2020 год]
Authors: T.M. Tatarnikova, S.Yu. Stepanov , Ya.A. Petrov , A.Yu. Sidorenko
Visitors: 925
The GIS usage is necessary for effective solution in scientific, practical and theoretical problems of in-ventory, analysis, modeling, forecasting, environmental system management, community territorial or-ganization. The relevance of this paper is due to the need to improve the methodological apparatus for detect-ing possible threats in the context of their growth dynamics and changing the impact concepts on spa-tial data in GIS for decision support. In the course of scientific paper, the requirements to the protection system structure, there is analyze during information processing in geographic information systems. The article presents a method for solving problems of creating and supporting the spatial information protection system operation in a GIS. To solve this problem, the authors selected an artificial neural network algorithm and modified to detect distributed DDoS attacks, the purpose of which is «service denial» and prevent legitimate users from accessing the attacked application. Based on the presented algorithm, the authors developed a program in a high-level language – Py-thon. This program includes a component number responsible for: replenishing the knowledge base of a neural network, which in turn allows you to build arbitrary ANN architectures; packet traffic analyz-er, the so-called sniffer, which provides packet filtering according to certain network protocols of the OSI model; a connecting module that allows you to send sniffer data to the ANN knowledge base. The resulting neural network can operate in two training modes: without teacher (self-taught), with teacher, which in turn allows the user to set the initial weights, or specify a file with a ready-made knowledge base.

8. Software module for automated calculation parameters for on-board electronic protection screens radiation protection equipment [№2 за 2020 год]
Authors: Zinchenko L.A., V.V. Kazakov , A.A. Mironov , A.V. Dorofeev , S.S. Kobylkin
Visitors: 823
When designing screens to protect onboard electronic equipment from radiation exposure, we need to perform a large number of calculations. The existing methodological base for calculating such screens does not allow performing automated analysis and processing of data obtained as a result of calcula-tions, and the designer has to do this manually. The article highlights the developed software module features that allow you to automate the calcu-lating process for the electronic equipment safety shield parameters from radiation exposure. The developed software module allows you to automate the process for calculating the effects of heavy charged particles on the radiation protection screen and evaluate the resistance of various design solutions to the effects of heavy charged particles. The article offers a client-server architecture that can be used to organize the multithreaded calcula-tion of safety shields on multiple connected clients. There is a database for storing already calculated screen parameters with online access. There is a principle of operation of the entire system in detail, as well as options for use cases. The authors have developed a web interface that allows you to enter data for calculations from any devices that have Internet access, as well as view the calculation results that have already been performed. Testing was on servers where the developed software module showed stable performance.

9. The pedipulator control system development for anthropomorphic robot AR-601M [№2 за 2020 год]
Authors: M.V. Tarachkov , O.V. Tolstel , Kalabin A.L.
Visitors: 824
The paper presents the pedipulator control system implementation for the anthropomorphic robot AR-601M manufactured by NPO Android Technika LLC (Magnitogorsk, Russia). The proposed control system consists of a driver that provides interaction between the on-board computer and the robot main microcontroller, and a control program that solves the inverse kinematic problem and allows planning the each pedipulator movements to a given point. The control program also has a graphical interface for displaying a three-dimensional model of the robot in space. To build this pedipulator control system, we used the freely distributed software platform Robot Operating System (ROS) and, in particular, ROS Control packages for implementing low-level interac-tion, MoveIt! for planning movements, RViz for visualization. Therefore, you must use the Linux Ub-untu 16.04 operating system. A program set for controlling pedipulators is in the C ++ programming language. The programming language choice of is due to the fact that to solve this problem, you need the best performance and a commensurate amount of cost for writing code. For the C ++ language, there is an ASIO input/output library that helps you interact with the robot controller. There are the experiment results on the proposed control system usage for the anthropomorphic ro-bot AR-601M pedipulators, as well as possible options for using the program described in the paper. The developed control system for pedipulators creates an abstraction level between hardware and software, which allows researchers to concentrate on solving the walking problem. In addition, the pro-gram has a convenient graphical visualizer. The paper considers the program updating possibility to use an inertial measurement module in its composition, which will allow obtaining data on the accelerations and angular velocities of the robot, and after applying the Kalman or Majvik filter to them, the orientation angles in space.

10. Decision support system for risk management of dangerous situations in complex gas supply systems [№2 за 2020 год]
Authors: R.R. Kantyukov , Butusov O.B., Meshalkin V.P, V.M. Panarin
Visitors: 787
The authors developed an architecture, software and decision support system (DSS) for analyzing and emergency risk reduction in complex gas supply systems (CGSS) and for reducing the impact on forest areas. They use special modules for calculating technological and environmental risks in order to produce decisions in DSS. The ecological damage calculation is based on assessing the impact on the forest are-as by the emergency situations in the CGSS. During the development of the DSS software, there was a detailed classification of possible emergency situations in the CGSS, as well as special procedures de-veloped to produce appropriate system decisions for the dangerous situation control. When developing algorithmic and software for DSS, the authors introduced a number of new CGSS safety indicators for the first time. Technogenic risks (industrial, environmental, social) of the dangerous (emergency) situations on the CGSS were taken as the main risk indicators. Risks make it possible to assess probable environmen-tal damage in case of emergencies in the CGSS. Thepaper proposes to use segmentation of forest satel-lite images to assess the environmental damage caused by accidents in the CGSS. At the same time, computer modeling of atmospheric transport for gas emissions, biogeocenotic databases, and the ex-posure-response dependencies make it possible to determine the total damage to forests by the values of the exposure directly. While developing DSS software, the authors used Microsoft Visual Studio (C++ and C-sharp pro-gramming languages), Matlab package, as well as databases and modern means of information and computer technologies.

| 1 | 2 | 3 | Next →