Software & Systems

Articles of journal № 3 at 2020 year.

Order result by: Public date | Title | Authors |

1. Parallel processing for software system performance optimization [№3 за 2020 год]
Author: I.Yu. Artemov
Visitors: 3709
The article describes the supposition and working results to improve the operating time of one of the subsystems of the automation system of data harvesting procedure by operators that are out of the of-fice for EPR-system. There is information about the current systems which are dedicated to data har-vesting with mobile devices. The author determines the system weakness which works in the mode of interactive (online) or batch (offline) data exchange between the server part of the system and its mo-bility part, and proposes its remedies for, based on the connection of two ways of exchange data (online and offline). The author devotes the main attention to the mechanism description for generating data necessary and sufficient for the mobile device operation, on the basis of which it was developed software. During the operation of the software, the problems of the proposed scheme were identified, due to the unsatis-factory efficiency of the generated data. The author provides a detailed description of the data pro-cessing options. The author shows their organization’s capability in the instruction pipeline of data processing. The paper gives reasons for the need for work to optimize data processing procedures. The author proposes ways for speeding up existing processing procedures by using the capabilities of modern server systems, which are based on clusters of servers with a large number of CPU cores. The author provides analytical calculations of theoretically possible performance gains when using parallel data processing, as well as the results of testing the developed algorithm based on real data. At the same time, the actual performance increase of the algorithm was 2.38 times, while the theoretically possible 2.51 times.

2. Experimental investigation of effectiveness of algorithms for minimizing BDD representations of Boolean function syst [№3 за 2020 год]
Authors: Bibilo, P.N., Yu.Yu. Lankevich
Visitors: 5622
The mathematical apparatus of BDD is used in various fields of science. In the computer-aided design field, BDD allowed to obtain significant success in a formal verification of algorithmic descriptions of digital circuits. Design systems of digital VLSI use programs of BDD minimization at the stage of tech-nologically independent optimization. Many articles consider optimization of BDD representations of systems of completely defined Boolean functions. Main attention was paid to finding an arrangement of variables for minimizing the BDD complexity. The variable arrangement is used to decompose the initial functions of the system and sub-functions (cofactors), which are obtained in the process of de-composition. The complexity of a BDD is the number of nodes in it. Each node of the BDD corresponds to a complete or reduced form of Shannon expansion. Domestic CAD and logic optimization systems use several programs for minimization of BDD rep-resentation of Boolean function systems that implement various algorithms. The purpose of this paper is to study the efficiency of these programs for synthesis of combinational circuits from library CMOS elements. After obtaining BDD minimized as for the number of graph nodes and defined as a set of in-terconnected formulas of Shannon expansion, the synthesis of a logic circuit is performed in the same design library of digital CMOS VLSI; the results are compared by square and delay. In many cases, it is possible to achieve additional reduction of logic description complexity by performing additional logic minimization based on Boolean nets. In this case, the optimization criterion is the number of nodes in the Boolean net, without considering inversion of Boolean variables. It is agreed with “the number of literals” criterion in optimization of multi-level logic circuits. The results of experiments on standard examples are presented.

3. A web-based content generation system for a common digital space of scientific knowledge [№3 за 2020 год]
Authors: S.A. Vlasova, N.E. Kalenov , A.N. Sotnikov
Visitors: 6656
The paper discusses the web-based software package "Expertise", designed for expert evaluation of various types of objects to be included in the common digital space of scientific knowledge. Electronic publications, catalogues of library collections, databases, and information systems related to a specific scientific area, and other types of resources can be evaluated as objects. The assessment of objects ex-ercises during the session, which is the examination process, characterized by a given time period and a specific set of experts and objects. The software package consists of two modules: administrative and user-defined, and supports ex-pert and object databases. The administrative complex provides system setting for a specific session (creating metadata pro-files of experts and objects, rating tables, and the text of emails sent automatically to experts); entering and editing attributes of experts and objects in accordance with the established metadata profiles; viewing ratings assigned to objects; forming of rating lists of evaluated objects. As for the user-defined block, it focuses on the work of authorized experts. The expert evaluates an object by selecting one assessment from the normalized table. Each assessment consists of a pair of el-ements – a verbal expression and a corresponding numeric one. The user block interface allows to each expert to change the assessment previously assigned to the object, if necessary. The paper provides a detailed description of the “Expertise” complex functionality and its working technology illustrated by specific examples. The software package was tested in 2019 at two all-Russian competitions – for the best scientific monograph and for the best student qualification work. There were several hundred works at the competitions, and several dozen experts from different re-gions of the country participated in their evaluation.

4. Application of transfer learning for semiotic models to the foraging problem with real robots [№3 за 2020 год]
Authors: V.V. Vorobev , M.A. Rovbo
Visitors: 5407
The paper considers the problem of applying a transfer learning algorithm for agents with semiotic models of the world to the foraging task with real robots. The robot needs to collect randomly placed food items, which when collected appear in a new random place within the polygon. The mobile robot is controlled by an agent with a model of the world that describes sensor readings as predicates. The agent makes decisions based on a state-action value estimation table for Q-learning. The agent is pre-trained on a simplified model environment with discrete states in which actions are performed with a guaranteed deterministic outcome. In a real environment and its model, taking into account physics, actions can be performed incor-rectly due to a scheduler error, localization errors, and other problems, and the data analysis of sensor information gathered from the continuous world determines the environmental state. The authors show the implementability of the corresponding interfaces and portability of the con-cept from a simplified model environment both to its more complete model that takes into account physics and a real robot. The transfer learning application is successful, but the final performance of the agent is reduced (probably due to the incorrect assumption of the determinism of the world in a real environment) and the robot needs additional learning after the transfer. Gazebo was used as a simulator that takes physics into account while the real polygon was equipped with special markers and cameras for localization. The authors also used elements of augmented reality in the form of a virtual food module.

5. Data structures and the Quine–McCluskey method modification for minimizing normal forms [№3 за 2020 год]
Authors: Gdansky N.I., A.A. Denisov , Kulikova N.L.
Visitors: 4469
Logical methods of analysis and synthesis of systems of various nature are usually based on the use of descriptions of their structures and processes in them in the form of Boolean functions, which are equivalently reduced to conjunctive normal forms to unify the representation. For initial systems and processes, as a rule, the main criterion for optimality is the minimum number of components that make up their components, which simplifies the structure, reduces cost and increases reliability, so for model conjunctive normal forms, the problem of minimizing them is of great practical importance. The algorithm’s efficiency that processes complex objects (including conjunctive normal forms) significantly depends on the main and auxiliary data structures used to represent these objects. There-fore, based on the analysis of existing structures, a new complex three-level data structure for repre-senting conjunctive normal forms has been developed. The lower level of the entire combined structure is the data structure for a single clause. Lists of clauses with the same number of letters form the middle level. An array of lists ordered by the lengths of their clauses sets the entire conjunctive normal form at the top level. The using lists, pointers to them and single elements, and the ordering of clauses by lengths make it possible to radically reduce operations for rewriting information and ordering it in the process of con-verting conjunctive normal forms. Based on the developed complex data structure, the authors developed a modification of the well–known Quine-McCluskey method used to reduce perfect normal forms. The combined use of the proposed data structure and the modified method makes it possible to sig-nificantly reduce the total number of operations while minimizing conjunctive normal forms compared to the basic version of the Quine–McCluskey method. This is achieved by reducing data re-processing, and by using special logical conditions, an addi-tional reduction is achieved in the total number of checks of letters in clauses when comparing them.

6. Agent-based models, methods and tools for the development of compatible problem-solvers for intelligent systems [№3 за 2020 год]
Authors: V.V. Golenkov, D.V. Shunkevich
Visitors: 6068
The article is devoted to the development of agent-based models, methods, and tools for the develop-ment of compatible problem-solvers for intelligent systems that can solve complex problems. We con-sider the requirements for such solvers, a model of a problem solver that meets the requirements, as well as the methodology and development tool and modifying such solvers. The main problem considered in the paper is the problem of low consistency of the principles un-derlying the implementation of various models for solving problems. As a result, the simultaneous use of different models for solving problems in a single system when solving the same complex problem is substantially difficult, it is practically impossible to reuse technical solutions implemented in any sys-tem, in addition, there are practically no complex methods and tools for developing problem solvers capable of providing the solver development process at all stages. It is proposed to use a multi-agent approach as a basis for solving the problem of compatibility of problem solvers. The process of solving any problem is proposed to be divided into logically atomic actions, which will ensure compatibility and modifiability of the developed solvers. It is proposed to consider the solver as a hierarchical system consisting of several interconnected levels, which provides the ability of the independent design, debugging, and verification of components at different levels. The proposed models, methods, and tools were applied in the development of a number of proto-types of intelligent educational systems, as well as a prototype of a recipe production automation sys-tem.

7. Semantic models and the method of coordinated development of knowledge bases [№3 за 2020 год]
Authors: N.A. Gulyakina, I.T. Davydenko
Visitors: 4517
The paper discusses an approach to easily modifiable hybrid knowledge bases creating based on se-mantic networks with basic set-theoretic interpretation. The paper proposes a semantic model of knowledge bases, including a set of top-level ontologies that ensure the compatibility of various types of knowledge. A distinctive feature of the proposed model is the use as a basic element of the model of not atomic elements of the semantic network, but signs of knowledge base fragments, called structures. The pro-posed model provides the consistency of different types of knowledge within the knowledge base, as well as the ability to knowledge bases structuring according to an arbitrary set of features. In addition, the paper describes a method for knowledge bases development based on this model, focused on the concerted development of a knowledge base by a distributed team of developers. A dis-tinctive feature of the method is its focus on the reuse of previously developed knowledge bases com-ponents of various complexity, as well as the presence of a formal ontology that describes the activities of the developers of knowledge bases in accordance with the method. The proposed method is implemented in the form of a system for the collective development of knowledge bases, which is embedded as a typical subsystem in each developed system and thus pro-vides the possibility of developing a knowledge base directly during its operation. The use of the proposed models, methods, and tools allows ensuring semantic compatibility of vari-ous fragments included in the knowledge base, as well as reducing the time spent on the development of knowledge bases.

8. Design and programming features when creating information systems [№3 за 2020 год]
Author: R.D. Gutgarts
Visitors: 6186
In the market of functional software designed to automate various problems and management levels at enterprises and organizations, a large number of ready-made design solutions are presented. Such ap-plications, as a rule, are positioned as typical ones and provide adaptation to specific conditions at the automation object by adjusting special parameters and (or) a relatively small amount of additional pro-gramming. However, the number of economic entities is large. All of them differ in production capacities, product range, and other characteristics. In a market economy, each of them is interested in maintaining and developing their competitive advantages. Therefore, it is not always possible to find a suitable au-tomated information system (IS) that satisfies all the necessary requirements. And then there is a need to create a system in accordance with an individual order. In such cases, its quality will completely de-pend on the correctness of the formulated requirements on the part of the customer and the profession-alism of the developer. The requirements are reduced to algorithms for solving problems, and the algorithms themselves are defined and agreed upon at the design stage of the information system. When programming, only cod-ing of algorithms is performed, i.e. recording them in the form of formal constructions in an algorithmic language suitable for execution on a computer. The paper shows the real dependence of the programming phase on the design phase since pro-gramming is always secondary to design. An exception is the technology used for rapid application de-velopment when an inextricable link is formed between the two stages and they are essentially com-bined into one. The importance of design is proved and some of its features are considered. To illustrate, examples of private design solutions are provided.

9. Text analysis method for tariff classification goods in customs [№3 за 2020 год]
Authors: E.V. Zhiryaeva , V.N. Naumov
Visitors: 3814
The use of artificial intelligence in customs administration is the most important practical sphere of the digital transformation of socio-economic systems. The paper considers one of the particular problems of this area: the choice of a product code according to the tariff classification based on declarations presented in Russian. The object of study was a voltage stabilizer. Analysis of customs declarations, performed by a person using the keywords in the product description, showed the need for machine learning methods. To do this, 1005 customs declarations were submitted, filed for three commodity items, which were regarded as three classes in the classification problem. Using the Orange Anaconda Navigator platform, it was possible to apply visual design methods to build a workflow diagram for solving the problem. The diagram includes a pre-processing stage, at which word clouds and a word bag were built, and a da-ta set was formed, the columns of which are lemmas, and the lines are individual declarations. In order to reduce the dimension of the problem, methods of filtering, removing n-grams, and stop words were applied. The resulting data set allows us to choose the best classifier in terms of accuracy, specificity, sensitivity, as well as using the error matrix and AUC curve. Training and test samples, as well as cross-validation were used.. The best in terms of the range of indicators analyzed was the classifier based on logistic regression, the equation of which made it possible to determine the most important lemmas for solving the classification problem. Since the complexity of solving the problem depends on the number of identifiable classes, it is ad-visable to use specific classifications for a small number of classes, including them in the information-analytical systems along with accounting systems, databases of customs declarations, request-response systems, and others.

10. Software for collecting, processing and transmission of technical condition data surface of the electric motor collector [№3 за 2020 год]
Authors: A.L. Zolkin , V.S. Tormozov , T.N. Bushtruk, M.V. Petrushova
Visitors: 3851
The paper explores a complex methodology and presents the results of developing a software product for monitoring the wear of collector plates of electric locomotives in the conditions of repair produc-tion using modern information technologies. The authors offer a software product that allows you to quickly and with a high degree of accuracy to calculate the wear parameters of the collector plates of electric locomotives, depending on their mileage. The low reliability of locomotive electric motors leads to the appearance of failures in the way. At the same time, damage to electric motors leads to the greatest expenditure of time and money for un-scheduled repairs, as well as the greatest downtime of trains on the line. One of the most difficult to manufacture and responsible for the operation of locomotive electric motor units is the collector. Its surface reflects the presence of hidden defects and violations of the operating mode of the electric mo-tor. Thus, the information technologies using in the repair and testing of locomotive electric motors, the electronic database creation, and resource forecasting in order to optimize the terms of trouble-free op-eration are now becoming relevant. When repairing electric motors in locomotive depots, it is proposed to use software that will im-prove control over the quality of this node repair. The introduction of software intended for the collec-tion, processing, and transmission of data on the technical condition of the surface of the electric motor collector in motive-power depots will provide a number of positive, but economically difficult to con-sider factors into the repair industry, including increased discipline and work culture, the need for ac-curate adherence of measurement technology.

| 1 | 2 | 3 | Next → ►