The article considers a classification problem of the interfaces for the transport infrastructure of the Smart City. The status of this topic in scientific research is reviewed, including existing approaches to interface classifications. The main elements of the transport infrastructure are highlighted, namely the following: infrastructure, autonomous vehicle, and environment (which is additionally divided into subjects and objects). The scheme of interaction of elements is described and the location of interfaces on it is indicated. The classification approach uses a combination of directional information exchange between elements. As a result of solving the problem, 9 classes of interfaces are distinguished, which have the necessity and sufficiency. A description of each interface and its practical example are given. In addition, the interface classes are written in a formal form. A direction for further research is proposed.

The implementation of the ternary balanced number system (TBNS) for fractional numbers in scientific notation is considered. Graphical representations of fractions with excess/deficiency are given. In the program model of the ternary coprocessor, the accumulator register of rounding errors is added. The algorithm of the accumulator register operation is formulated. The character 7 is used for encoding –1. Examples of some numbers are given in the form of fractions, in decimal representation, and in the TBNS form with excess and deficiency. The authors consider in detail the procedure of numerical integration for function values in the TBNS form with a deficiency by the rectangle method. For six addition operations, the error-compensating value is accumulated twice in the storage register. The final result is the same as the expected one, up to the last digit. In the second

The paper continues the study in the area of intelligent photovoltaic array maximum power point tracking (MPPT) systems under non-uniform insolation. We developed, tested, and logically designed the MPPT system of the photoelectric array based on a modified fuzzy neural network as software. The results of comparative experimental simulation demonstrate that the developed intelligent system achieves competitive performance and robustness, as compared to a classical MPPT system based on the perturbation and observation algorithm, or particle swarm optimization under non-uniform and uniform insolation. The modular architecture of the software “System for tracking the maximum power point of a photovoltaic array based on a modified fuzzy neural network” and its multi-level class hierarchy by UML are developed.

The article continues the authors’ research in the field of E-networks and construction of mathematical models of complex technical systems with adaptive structural and parametric configuration. The article presents an E-network approach to the construction of artificial neural networks using the procedure of reconfiguration of the topology of the simulated network. This solution allows presenting the artificial neural network model in a more compact form and solves the problem of its interfacing with other components of the model. The resulting E-network mathematical model of artificial neural network is capable of solving a large class of problems, including identification of the behavior of a human operator (a group of operators) in automated process control systems.

The article describes the preparation of input parameters of the expert system for the construction phase of production rules. The analysis of the possibility of dimensional reduction of input parameters by factor analysis methods such as principal component analysis and factor analysis is made. The adequacy assessment of the general sample for constructing a factor model is verified by Bartlett's test and the Kaiser-Meyer-Olkin Test. The model is subjected to qualitative assessments: the proportion of explained variance, the generality of factors, and the interpretability of the model. Based on the obtained factors, correlation graphs with input parameters are plotted, and graphically represented as a map of relationships. As a result, a model consisting of two factors is obtained, which indicates that the input parameters of the system have a complex correlation and cannot be reduced.

The article studies the performance of algorithms for generating random numbers from the gamma distribution. In the simulation model of the column jet-emulsion reactor, the gamma distribution is used to create particles for such parameters as the size and composition of substances in the particle. ActionScript 3.0 is chosen as the programming language. Seven algorithms for generating the gamma distribution are implemented: Marsaglia and Tsang, Cheng and Fist (two versions), Ahrens and Dieter, Tanizaki, Schmeiser (two versions) with a parameter value α > 1. To implement the Marsaglia and Tsang algorithm, algorithms for generating a normal law of distribution are considered: Box-Muller, Marsaglia-Bray, Devroy, central limit theorem, Neumann, the ziggurat method. The comparison of the generation of the gamma distribution in two versions is made: without preliminary initialization of the initial values and with it. The fastest for 1 < α < 2 was the Cheng and Fist algorithm. For α > 2 Marsaglia and Tsang's algorithm is stable with increasing gamma distribution parameter α.

The paper solves the problem of verification of the physical model (experimental setup) of the vertical electrical sounding method relative to the mathematical modeling of this method in the IPI2Win program when studying two-layer geoelectric sections and interpreting the results. The work aims to identify the limits of applicability of the physical model and its correlation with the mathematical model of the IPI2Win program, increase the accuracy of the modeling, and verification the obtained results using the mathematical model. Several parameters of the physical model are obtained, which impose restrictions on the implementation of the physical model of the vertical electrical sounding method as well as the use of a mathematical model of the IPI2Win program for processing the results by the physical model.