Journal influence
Higher Attestation Commission (VAK) - К1 quartile
Russian Science Citation Index (RSCI)
Bookmark
Next issue
№4
Publication date:
09 September 2024
Latest issue articles
Order result by: Public date | Title | Authors |
1. Physical model of a robot manipulator: Optimization of real-time intelligent control processes using a genetic algorithm [№3 за год]
Authors: Katulin, M.S., A.G. Reshetnikov, Ryabov, A.R., Ulyanov, S.V.
Visitors: 784
When developing an intelligent control system, it becomes necessary to select the optimal parameters for the system to achieve a management goal. A selection of system parameters involves using a mathematical model or similar methods. This paper considers the problem of configuring an intelligent control system with undefined (not taken into account in the mathematical model) parameters. As an example of such system, the paper uses a control system for a four-link robot manipulator in a feedback loop based on a machine vision system. This example clearly demonstrates the operation of a control system characterized by external factors that affect system behavior and that are not always taken into account. In the example under consideration, the external factors are drive backlashes, errors/inaccuracies during the operation of a machine vision system, or a sudden change in a target position. The work describes in detail the robot manipulator design, a machine vision system and the controllers used. The first step of configuring the system involves empirical selection of the desired parameters, then the authors perform optimization using a genetic algorithm. The paper presents a method of applying a genetic algorithm to optimize the control system parameters on a physical object in real time. The paper considers algorithmic features of the applied genetic algorithm, including the features of implementing the fitness function. There are optimization results and a comparative analysis of the system operation with different parameters. The paper demonstrates the possibility of implementing such systems and using genetic algorithms in real time. The presented methods of intelligent systems design technology can be transferred to other control objects.
2. Applying the feature selective validation method to compare experimental or simulated datasets [№3 за год]
Authors: Shaymanov, N.Yu., Avraamov, V.P., Ivanov, A.A., Kuksenko, S.P.
Visitors: 692
This paper is intended to introduce domestic engineers and researchers to the feature selective validation (FSV) method, which is widely used by foreign specialists in radio-engineering and electronics for quantitative evaluation of the agreement between two datasets (for example, when comparing the results of simulation and experiment). For this purpose, the paper presents a new open source software tool based on the FSV method. It also gives theoretical information about the method and shows examples of its use, describes a FSV data comparison procedure in detail. The paper notes the peculiarities of data partitioning into subranges, the basics of calculating measures indicating the difference between sets, and the interpretation of calculated measures when constructing evaluation histograms. The developed block diagram visually describes the algorithm of the FSV method and the specifics of its software implementation. A software tool based on the FSV method was developed in C++ using this block diagram. The paper provides information about the development tools, software architecture, graphical user interface, and functionalities. The final part of the paper presents examples of applying the FSV method in the form of global difference measure histograms obtained by comparing four data sets describing complex and non-linear frequency dependencies: voltage in a multi-conductor transmission line, shielding effectiveness of a metal enclosure, modulus of the transmission coefficient for a wire structure, electric field strength outside an apertured cavity. The results showed that the developed software tool and the FSV method allow obtaining a reliable quantitative assessment of the agreement between two data sets, even when their qualitative visual comparison is difficult.
3. Identifying Pc3 type geomagnetic pulsations from INTERMAGNET data: MATLAB Big Data methods [№3 за год]
Author: Korobeynikov, A.G.
Visitors: 718
MATLAB system is widely used in solving fundamental and applied problems in various subject areas. This work describes using Big Data technology methods implemented in MATLAB to solve the problem of identifying Pc3 type geomagnetic pulsations from data supplied by magnetic observatories included in the international INTERMAGNET network. The solution involves passing the preprocessed data set through a designed infinite pulse bandpass digital filter (infinite impulse response filter – IIR filter) and then removing the nonlinear phase distortion of the filter. The IIR filter is designed using the Zolotarev–Kauer method. The filter parameters are based on the period range of Pc3 geomagnetic pulsations 10÷45 seconds. The choice of a design method is determined by the developed IIR filter characteristics: 1) a balanced behavior of frequency response ripples both in the passband and delay band; 2) relaxing the requirements for the frequency response shape in the delay band that allows a quick transition between the passband and the delay band; 3) the order of the IIR filter designed using the Zolotarev digital filter method and its complexity are minimal. One of the requirements when solving the indicated problem is the condition of using Big Data technologies implemented in MATLAB. This requirement arises from a large amount of initial data: continuous recording of data on the geomagnetic field state through four channels for a whole year with a sampling frequency equal to one second. The obtained results can be used to solve both applied problems, for example, analyzing the space weather state, and fundamental ones, for example, further development of the theory of geomagnetic pulsations. The IIR bandpass filter design and calculations were carried out in MATLAB R2023b.
4. Software implementation of algorithms for knowledge base prototyping with visual modelling and transformations [№3 за год]
Authors: Dorodnykh N.O., Yurin A.Yu.
Visitors: 833
The paper describes a special purpose web-oriented software system called Knowledge Modeling System (KMS) designed for experts and system analysts. The system enables building visual conceptual models in the form of state transition diagrams, event and failure trees, as well as automatic generation of knowledge base code based on their transformation. The obtained knowledge bases can become prototypes in further development of intelligent systems. The created system bases on the principles of visual modelling and model transformations. The latter principle enables describing the correspondence between the elements of different models (notations) and their transformation. Forming transformation operators for conceptual models involves describing their models (metamodels) that include basic elements and relations. The paper shows transformations as a set of correspondences between model elements, they are implemented in the general-purpose language PHP. CLIPS and OWL2 DL are the target languages. The paper describes the developed system, including the method of creating knowledge bases using the transformation of conceptual models, main functions and architecture. The implemented method includes the stages of building a subject area conceptual model, XML representation of conceptual models, analysis of XML-structure of the model, formation of ontology model or products, modification of the obtained knowledge, generation of knowledge base code in the target language. The authors of the paper present examples of using KMS in terms of creating knowledge bases in the field of technological safety. The examples show how to solve the problems of diagnosing and forecasting the technical state of objects and systems for describing accident dynamics when oil flows out of the reservoir, as well as planning the failure analysis algorithm.
5. Cyber-physical system objects: Organizing operation data storage [№3 за год]
Authors: Korostelyov, D.A., Podvesovskii, A.G., Zakharova, A.A.
Visitors: 711
The paper presents the results of research in the field of organization and software support when storing operation data from cyber-physical system objects obtained during experiments. The identified features of their storage are the following: special data formats, large amounts of data with different structuring degrees, supporting universal structured data formats, storing information about the structure and scenarios of experiments. The authors of the paper have analyzed technologies for building storage systems and special-purpose data formats that consider these features. They also proposed an approach to building a software system for storing and preprocessing experimental information about the functioning of cyber-physical system objects. The architecture and peculiarities of the system implementation are considered on the example of storing data of experiments with groups of jointly operating unmanned mobile vehicles when solving tasks of moving in space and transporting cargoes. The architecture assumes a server application and a web-client interacting with it by means of an API-interface based on REST principles. The server application also includes a subsystem for preprocessing experimental results and a data source format converter. The paper proposes to use a two-level structure of the data model for storing the experimental information. The upper level is intended for storing information about the structure and parameters of experiments and is implemented as a database in the PostgreSQL system. The lower level stores streaming information received from primary data sources. This level is implemented as a cloud storage built based on Yandex Cloud using S3 technology. The results of testing the developed system in real experiments confirmed the validity of the selected architectural solutions. Practical significance of the performed researches consists in developing approaches to the construction of scalable software systems for storing operating data of cyber-physical system objects. This contributes to organizing tasks and control scenarios of these objects when developing new control technologies and tools.
6. MATLAB/GNU Octave for calculating per-unit-length parameters of multiconductor transmission lines using the method of moments [№3 за год]
Authors: Maksimov, A.E., Snetkov, P.P., Ivanov, A.A., Kuksenko, S.P.
Visitors: 639
The paper describes a prototype of software for calculating (extraction) primary per-unit-length parameters (electrostatic and electromagnetic induction coefficient matrices) developed using MATLAB/GNU Octave programming language. The program extracts parameters of multiconductor transmission lines with arbitrary cross-section, any number of dielectric layers and conductors on each layer. It uses the numerical method of moments. There are implemented uniform and 2 methods of non-uniform segmentation of transmission line cross-section boundaries (corner and projection segmentation). There are also 3 methods of iterative segmentation (segmentation of all segments, only those segments whose length exceeds the set threshold value and segments with the highest charge density). Four methods of solving the matrix equations are implemented: LU-decomposition, Bl-BiCGStab, Bl-IDR(S) and Bl-GMRES. Parameter extracting is followed by their physicality check. The developed prototype is also able to perform multivariate analysis, i.e. automatic iterative extraction of transmission line parameters when the input data (geometric and electrophysical parameters) change according to a certain rule. The results of multivariate analysis can become a base for statistical analysis (calculation of basic statistical characteristics such as mathematical expectation, variance, standard deviation and confidence interval). Verification of the prototype showed that the extracted parameters agree with similar ones from other software. Great flexibility of configuration, modularity, different methods of segmentation and matrix equation solution allow using both the prototype as a whole and its separate modules for research and academic tasks. The calculated primary per-unit-length parameters of the line help to calculate its scattering parameters easily.
7. Modifying a GH-graph shortest path search algorithm for analyzing complex technical systems [№3 за год]
Author: Zyablova, E.R.
Visitors: 666
The paper proposes one of the approaches to modelling complex technical systems on the example of solving the problem of forming zones of influence of the extended perimeter security system objects. The approach uses the socalled GH-graph (fuzzy graph with different types of vertices and multiple and different types of links) and certain algorithmic tools. This paper follows author’s previous works that detail the possible graph model of the system (or its part), the algorithm of GH-graph proportional partitioning, and its application to solve the problem. This study includes synthesis of a modified Ford-Bellman algorithm for finding the shortest paths of a GH-graph. The modified algorithm allows analyzing different types of information flows in complex technical systems. It is characterized by the ability to find distances for complete or truncated sets of vertices and/or graph links, e.g. for given type links. There is a list representation of multiple and different types of GH-graph links. The formulated criteria for finding shortest paths in the GH-graph are the following: criterion 1 – selecting a type (types) of vertices involved in the algorithm; criterion 2 – selecting the types (vectors) of edges involved in the algorithm. To this end, it is possible to leave the complete sets of vertices and edges of the model in the graph list representation. The determined value of the computational complexity of the modified algorithm does not exceed the complexity value of the original Ford-Bellman algorithm. The operating time of the proposed algorithm is reduced due to using multiple edges in the GH-graph as a vector, which allows combining a number of different types of edges. The result of the modified algorithm is a distance matrix considering complete or truncated sets of vertices and/or edges for subsequent calculation of graph model metric characteristics by means determined by early research. The authors consider the proposed approaches to modeling complex technical systems in detail by determining the zones of influence of technical devices (quadrotors) on system objects and selecting a suitable quadrotor model. There is a brief description of the software implementation of the graph characteristics calculation module.
8. Basic principles of generalized regression neural network when filling missing values in datasets [№3 за год]
Authors: Tatarnikova, T.M., Bozhenko, V.V.
Visitors: 612
The paper discusses the relevance of filling missing values in the initial data set at the preprocessing stage when solving problems of data analysis and machine learning. The authors of the paper propose to use a generalized regression neural network to solve the problem of filling missing values in the initial data set. In comparison with the statistical method based on the mean or median value per column, it implies taking into account possible dependencies between data. The paper considers the basic principles of the generalized regression neural network, its architecture features, advantages and disadvantages. It also shows that the advantages of the generalized regression neural network include fast training on a small amount of input data and the ability to predict missing values due to its capability to approximate complex functions. The authors also give an algorithm for using a generalized regression neural network for gap recovery. The algorithm is one-pass; it adjusts the weights of links between network layers, a radial basis function parameter, and a learning rate during one-pass training of the neural network. Training the neural network aims to minimize the prediction error, which is RMS error. There is a scheme for filling in the missing values using a statistical method. The paper presents an algorithm for applying the omission filling scheme based on determining the average feature according to the available values, that is the data located above the feature column cell to be filled in. The prediction of missing values by the statistical method was also evaluated using the mean square error. The authors demonstrate the results of training the generalized regression neural network model and applying the statistical method on a validation dataset. Comparison of the results of filling in missing values by two methods showed the advantage of the generalized regression neural network on a significant (large) dataset.
9. Justifying the choice of a rational structure of the system under study: Functional modeling of two complex competing systems [№3 за год]
Authors: Dolgov, N.V., Ilin V.А.
Visitors: 600
Classical functional modelling is based on the structure analysis of links between individual functions of the modelled process. It does not take into account their interaction with other systems, including mutual influence of separate functions of two or more systems. The paper shows a general analysis of functional modelling of two complex interacting systems with antagonistic goals. When these systems interact, one of them purposefully attacks the other and the opposite side defends itself. It passively or actively counteracts these attacks, i.e. there is a certain competition between the systems. As an example, the paper considers a certain abstract process presented in the IDEF0 technology notation, which is a functional model of two complex interacting systems with antagonistic goals. Interpreting this model as a digraph forms a matrix of relations of the modelled process functions. This matrix is a basis for deriving mathematical dependencies for determining the effectiveness of the functional model of two complex interacting systems with antagonistic goals.
10. Cubic graph edge coloring in the problem of calculation parallelization on an unstructured surface computational grid [№3 за год]
Authors: Gulicheva, A.A., Rybakov, A.A.
Visitors: 651
The paper considers the issue of improving the performance of finite-volume numerical methods on shared-memory computing systems. In these methods, data conflicts are possible at the stage of calculation of conservative value flows across computational cell boundaries. This leads to performance degradation, especially with a large number of simultaneously operating flows. To eliminate data conflicts, the paper proposes a solution based on partitioning the set of processed cell boundaries of the computational grid into subsets without conflicts and processing these sets separately. The authors consider the solution on the example of calculations on surface unstructured computational grids. For them the problem is reduced to the problem of constructing a cubic graph edge coloring. To construct a cubic graph edge coloring, the authors apply two algorithms: the trivial linear algorithm of coloring in five colors and the algorithm of Tait coloring in three colors. The authors of the paper compare two coloring algorithms, as well as measure the influence of the proposed approach on the efficiency of calculation parallelization on a surface computational grid. The use of the data dependency elimination approach using cubic graph edge coloring was tested on the numerical problem of calculating body surface icing. Specialists performed launches on an Intel Xeon Phi microprocessor with a large number of parallel flows. The results showed that when the number of flows increases up to 144 and more, the efficiency of parallelization using edge coloring is twice as high as that of the usual approach of dependency elimination using OpenMP directives.
| 1 | 2 | Next → ►