Journal archive

In recent years, bioinspired algorithms based on the use of a population approach and a probabilistic search strategy have become especially popular among researchers involved in multidimensional and multicriteria optimization. Such algorithms are based on the principles of cooperative behavior of a decentralized self-organizing colony of living organisms (bees, ants, birds, etc.) to achieve certain goals (for example, to meet nutritional needs). However, their practical application encounters a number of difficulties leading to a decrease in convergence. This article discusses the possibility of modifying the artificial bee colony algorithm by using a hybridization strategy with various data mining methods. One of these difficulties is the lack of a reasonable approach to determining initial search positions. As a solution, it is proposed to divide the population into clusters, the centers of which will be the initial positions. The need for interaction between individuals makes it advisable to use fuzzy clustering, which allows the formation of intersecting clusters. Another difficulty is associated with the choice of “free” parameters, for which the authors have not developed recommendations for choosing their optimal values. To solve this problem, it is proposed to use the idea of coevolution, which consists in the parallel launch of several interacting subpopulations, for each of which different “settings” are applied. The proposed algorithm is applicable to multidimensional optimization tasks, in which it is necessary to find such a combination of different types of elements belonging to some “large” population that will ensure the achievement of the maximum effect under given restrictions. Examples of such tasks are determining the species and quantitative composition of plants to form the terrestrial ecosystem of a carbon farm or mass recruiting, which consists of selecting a large number of personnel for the same positions.

The results of research are presented, the purpose of which was to develop a method for predicting the outflow of clients of a commercial bank based on the use of machine learning models (including deep artificial neural networks) for processing client data, as well as the creation of software tools that implement this method. The object of the study is a commercial bank, and the subject of the study is its activities in the B2C segment, which includes commercial interaction between businesses and individuals. The relevance of the chosen area of research is determined by the increased activity of banks in the field of introducing digital services to reduce non-operating costs associated, in particular, with retaining clients, since the costs of attracting new ones are much higher than maintaining existing clients. The scientific novelty of the research results is the developed method for predicting the outflow of commercial bank clients, as well as the algorithm underlying the software that implements the proposed method. The proposed ensemble forecasting model is based on three classification algorithms: k-means, random forest and multilayer perceptron. To aggregate the outputs of individual models, it is proposed to use a learning tree of fuzzy inference systems of the Mamdani type. Training of the ensemble model is carried out in two stages: first, the listed three classifiers are trained, and then, based on the data obtained from their outputs, a tree of fuzzy inference systems is trained. The ensemble model in the proposed method implements a static version of the forecast, the results of which are used in a dynamic forecast performed in two versions – based on the recurrent least squares method and based on a convolutional neural network. Model experiments carried out on a synthetic dataset taken from the Kaggle website showed that the ensemble model has a higher quality of binary classification than each model individually.

Riding simulation machines allow training horse-back riders regardless of environmental and weather conditions, horse fatigue, stable remoteness and other factors. Riders can practice complex movements without fear of potential injury related to sometimes unpredictable behavior of the horse. This type of training machines requires special software for selecting the relevant action mode for the person during the rider’s practice. Thus, the selected behavior of the horse can be simulated and the injuries of horse-back rider can be avoided. For this purpose, the model of horse-back rider, taking into account its training level and the horse motion, is required. The model of “person – simulation machine” combination is proposed in the article. The proposed model allows developing operating modes taking into account the sportsman-training machine interaction. The mechanical model is an exosuit with a mobile pole, consisting of four links, the rider’s feet, shins, hips, and the corps are attached to. The pole corresponds to the mass center of the horse, the rider interacts with by standing on the stirrups. The pole motion is implemented by a telescopic link attached to immobile base. The relative rotations of the links are implemented by cylindrical hinges with negligible friction. The proposed model has been implemented as software in the “Wolfram Mathematica 11.3” environment. It has been designed for training machine dynamics simulation of the “horse ­– rider” system. The software includes several modules: 1) the module for specifying the structure of training machine mathematical model, generalized coordinates, and for auto-compilation of the corresponding system of differential equations; 2) the module for specifying the programmed model motion and calculation of the required control torques in the hinges; 3) the module for the Cauchy problem numerical solution; 4) the module for the animated visualizing of the model motion, and for exporting the obtained graphic results and numerical calculations. The developed software allows conducting dynamics analysis of mathematical model for the considered system based on the solution of both direct and inverse dynamics problems. Also it can be suggested for designing training machines with programmed operation mode. It has been shown that application of this software accelerates the development of training machines.

In the article, using computer modeling, an analysis of the operation of distance protection of backup transformers is carried out, providing alternative power supply for the own needs of a nuclear power plant through a backup busbar when disconnected from the power system. During the transition from the main network to the backup network, there is a short period of power off to the sections. At the same time, the powerful motor loads of the sections that have received power begin to operate in the freewheel braking mode. Then the reserve is automatically switched on, which can occur at a more or less favorable moment. At an unfavorable moment, self-starting may be electrically more severe than a short circuit on the busbar. When combining a new network from a backup auxiliary transformer and an autonomous circuit of running machines, transient processes of electromagnetic interaction between machines switching to generator mode and the new network arise. There are some optimal favorable moments for merging networks when it is advisable to carry out self-starting. Due to the complexity of the mathematical description with a high order system of equations and a large number of interaction objects, it is advisable to study these processes using computer modeling. A practical question, which is answered by the calculations and computer modeling, concerns the value of the distance protection setting for the backup transformer for auxiliary needs. Using the developed structural simulation model in MatLab, a series of experiments was carried out on the run-down and self-starting of auxiliary sections with their given composition and load. The range of self-start time corresponded to the operation of automatic switching on of the reserve sections by distance protection. Calculations and modeling show that the adopted setting can be adjusted, ensuring protection of a larger length of the busbar with possible options for connecting auxiliary sections to it. The model is supplemented with an add-on in the form of an external program for detailed processing and visualization of data obtained from oscilloscopes of the MatLab structural model.

Currently, the development of non-invasive diagnostic methods for electric machines and especially generators, as the main producers of electric energy, using magnetic field fixation sensors is an urgent scientific and technical task. To solve this problem, research has been conducted to develop a method for troubleshooting a synchronous generator by analyzing an external magnetic field. The article discusses the main issues of computer modeling of an external magnetic field in the FEMM 4.2 system for troubleshooting. Mathematical expressions are given for creating a computer model of a synchronous generator in two-dimensional space, on the basis of which a numerical solution of the external magnetic field is formed using the finite element method. The modeling takes into account the design features of the synchronous generator. The main stages of work in the FEMM 4.2 system are defined. The geometric model of the synchronous generator is imported from the computer-aided design system. The physical properties of all elements of the model are determined by the structural materials of the synchronous generator and the external space. The control program, created on the basis of the algorithm presented in the article, allows you to simulate the rotation of the inductor of a synchronous generator, automate the calculations of the electromagnetic field and display the results. An example of using a computer model of a synchronous generator for troubleshooting by examining an external magnetic field is given. Based on the results of a numerical solution of the external magnetic field, a harmonic analysis of the magnetic induction of a working synchronous generator was carried out. The article shows that a diagnostic sign of static eccentricity of a synchronous generator inductor is the appearance of even harmonics in the magnetic induction spectrum of an external magnetic field Based on the results obtained, the dependence of the growth of even harmonics on the magnitude of the displacement of the inductor of the synchronous generator is determined.

Mathematical models are the main tool for the layout design of linear asynchronous electric motors. Existing design models of linear asynchronous electric motors assume that the device is powered from a source of multiphase symmetrical positive sequence current, which excludes from consideration the asymmetry of the inductor phase currents and the associated reduction in traction force. This approach is not flexible enough, since the operation of linear asynchronous electric motors in a wide range of movement speeds assumes a functional relationship between the current asymmetry coefficient and the movement speed. The study developed a mathematical model for design purposes that implements the functional relationship between the current asymmetry coefficient and the speed of movement. The mathematical model represents 13 linear equations linking geometric dimensions, current loads, electromagnetic field characteristics and mechanical forces with the speed of the secondary element of the electric motor. The current load of an electric motor has a variable degree of ellipticity depending on the speed of movement. An algorithmic study of the mathematical model was carried out and corresponding software modeling tools were developed. The proposed program algorithm, containing two subroutines, is implemented using the Maple symbolic computing software package. In order to expand the range of users, the software model is additionally implemented in C++. In this case, the functional dependence of the asymmetry coefficient of the exciting current wave on the speed of movement is implemented in the form of an iterative procedure for the case of connecting the winding of the linear asynchronous electric motors inductor to a source of symmetrical three-phase voltage according to the “triangle” circuit. Using the example of a three-phase linear asynchronous electric motors, the electromagnetic and mechanical characteristics of the electric motor were simulated. It was stated that the influence of current asymmetry on the mechanical characteristics of the linear asynchronous electric motors is small and is expressed in the region of high speeds. This makes it possible, in design tasks for the layout of an electric motor, to use models that involve powering the linear asynchronous electric motors from a symmetrical current source or to introduce a fixed asymmetry coefficient into them.

The paper is devoted to the current problem of reducing the percentage of defective products produced at serial steelmaking plants. To study the patterns of defect formation, a neural network was created that predicts the formation of defects such as “crack” in castings produced by the open-hearth method. To train the neural network, statistical data on the chemical composition of ore raw materials and the corresponding values of the percentage of defects were used. The data was taken under the conditions of an ongoing serial production process, which led to a high degree of noise in the information both on the chemical composition of the ore raw material and on the mechanical properties based on the results of its heat treatment. Outliers of statistical information were detected and removed using the original author’s neural network technique. A neural network model for controlling thermophysical and chemical-energy-technological processes of thermal processing of ore raw materials was created on the basis of a perceptron-type neural network with sigmoid activation functions. By conducting virtual computer experiments on a neural network model, some important dependences of the probability of formation of the defects under study on the content of manganese, phosphorus, silicon, chromium and sulfur were identified. Based on the identified dependencies, practical recommendations have been developed to reduce the percentage of defects by adjusting the chemical composition of ore raw materials. Despite the relatively low accuracy of the developed neural network model, the application of the practical recommendations obtained made it possible to reduce the percentage of defective products manufactured in a large-scale production process by 2.51 times.

Difficulties in prognosis of human behavior are due to the complexity of our cognition, routinely breaking the boundaries of classical rationality. The paper solves this problem for the simplest kind of such “irrational” behavior in which a single binary decision is made in three related contexts. Subjective meanings of these contexts relative to the basis decision alternative are represented by three qubit states, borrowed from quantum theory. These states are bound together by linear superpositions, which encode semantic composition of the contexts in the subject’s mind. The resulting theory supplements classical probabilistic model with nonlinear interference factor, accounting for the “irrational”, emotionally-semantic side of intelligence. This model is built for different realizations of two classic experiments used to study behavioral irrationality: the prisoner’s dilemma and the two-stage gambling task. In 24 such realizations, the interference phase is shown to fall in a narrow range of values, encoding regularities of semantic composition of contexts. Extrapolation of this regularity to novel experiments allows using the model in prognostic mode. This possibility is tested on the task of probabilistic prediction of target decision based on the same probability in two other contexts. For the prisoner’s dilemma and the two-stage gambling task the such prognosis has relative errors of 9 and 11% respectively. The proposed approach allows for putting other quantum models of cognition and decision to predictive and interpretable use, whereas its principles also apply to modeling of decisions with larger sets of contexts and behavioral options. By formalizing a novel type of semantic regularities behind “irrational” thinking, models of the present type open prospects for empowering the existing means for socio-economical analytics and prediction.