PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems

Virtual analyzers (VA) are widely used in the tasks of quality control of technological processes. In many cases, there is a problem of the adequacy of the mathematical model used in the structure of VA on the entire space of states of the technological process. The work investigates the possibility of determining the existing stable states of the process (modes) based on historical data on technological parameters and choosing one of several models in the VA structure, depending on the current operating mode of the technological unit. Purpose: improving the accuracy of the virtual analyzer of residual moisture of KCl after the drying stage in a fluidized bed (FB) by introducing into its structure an algorithm for taking into account the operating mode of the technological unit. Results: the trends of technological parameters of the drying process were investigated, the parameters that determine the operating mode of the technological unit were identified, a cluster analysis of estimates of the mathematical expectations of the values of these parameters was carried out. As a result, two operating modes of the FB apparatus have been identified. For each of the found modes, a statistical model was trained, which determines the dependence of the humidity at the outlet of the FB apparatus on the values of the technological parameters of the process. A mode accounting algorithm has been developed, which identifies the mode depending on the current values of technological parameters and selects the model corresponding to the mode. A computational experiment has been carried out to evaluate and compare the operation of the VA without an algorithm for taking into account the mode in its structure and for the VA with the built-in algorithm for accounting for the mode and a number of statistical models. The use of the algorithm for accounting for the mode in the structure of the VA reduced the indicator of the mean square error of the result of the VA operation by 16 %, and, thus, significantly increased the accuracy of the VA operation. Practical significance: the results of the study were used in the development of a virtual analyzer of residual moisture KCl after the stage of drying in a FB dryer at a potash enterprise. They allow determining the operating modes of the FB apparatus and increasing the accuracy of determining the quality indicator of the drying process, which makes it possible to more accurately control the process. In the future, the developed approach can be used, for example, to determine the modes of the technological process, characterized by the need to take equipment for repair and optimize the maintenance schedule.

Programmable logic integrated circuits (FPGA) are currently one of the main hardware bases for the implementation of digital devices and systems. As they developed from the 80s of the twentieth century, when they contained only hundreds of logical elements, to the present day, when the "top" FPGAs contain hundreds of thousands and even millions of logical elements, they gradually introduced means of scaling functionality, speed, power consumption. For example HyperFlex, Tri-Gate transistors, Voltage Reduction Technology (VRT), Dynamic voltage and frequency scaling (DVFS), Sleep modes, Power management. Concept of the energy modulated computing proposed by Alex Yakovlev. S.F. Tyurin is proposed the concept of scaling logical bases. Embedded diagnostic standards are being introduced. Now FPGAs are reaching an even higher level: they implement hardware accelerators for tasks traditionally solved in software. However, reliability scaling has not yet been fully achieved. Purpose development of the concept and theoretical foundations for scaling the reliability of FPGAs. Methods: analysis of the levels of ensuring the reliability of FPGAs, development of the concept of scaling the reliability of FPGAs, synthesis of a scalable architecture of FPGA logic. Results: the concept of scaling the reliability of the FPGA logic, the method of analyzing the levels of ensuring the reliability of the FPGA logic, the method of synthesizing the scalable redundancy of the FPGA logic. Practical significance: The developed theoretical foundations for scaling the reliability of the FPGA logic can provide the creation of a new, promising FPGA class.

The object of the presented research is the production process of the MK2000 utility vehicle in the conditions of the assembly shop of PJSC MZiK. The MK2000 is a technically complex product that is produced together with ten other types of technically complex products within the same workshop. The uncertainty factor in the production process is the unknown duration of the test cycle that each product is subjected to after the end of assembly. The purpose of the study: development of a method for calculating the duration of the production cycle of a product under conditions of uncertainty. Evaluation of the possibility of production planning and calculation of the annual schedule of shipment of finished products to the consumer by predicting the duration of the cycle. Methods: to solve the problem of predicting the duration of the production cycle, the dependence between the number of products of all types already in the process of testing at the time of the transfer of the product under consideration for testing, and the duration of the test of the transferred product is used. Since the production of various types of products is carried out at the same facilities, the regularity of the mutual influence of the production processes of various types of products is considered as the main one. The duration is calculated using a mathematical model obtained as a result of training the Random Forest algorithm. Results: a new method has been developed for calculating the duration of the production cycle of technically complex products for small-scale production enterprises based on statistical analysis of the results of the previous production activity of the enterprise. The pre-trained Random Forest machine learning algorithm was used for the calculation. Evaluation of the calculation results obtained using a mathematical model of the production process suggests that the prediction of the duration of the production cycle of products with an unknown duration of the test stage is possible using machine learning methods, and the accuracy of the forecast depends on the chosen method and the size of the training sample. Practical significance: the proposed method allows us to predict the duration of production and testing of technically complex products under conditions of uncertainty and can be used as a tool for calculating the annual schedule of product shipment to the consumer.

Internationalization and international activities are one of the priority tasks of the functioning and development of modern universities. Internationalization strategies or roadmaps for the development of international activities now play an important role in many universities, as well as in the general context of higher education policy. The need to find methods for determining the level of internationalization of universities arises in connection with the increase in global budgets for academic international mobility and international projects. From another point of view, until today, this area has been shown to be barely measured using scientifically based and comparable indicators. Purpose: The purpose of the study is to improve the efficiency of the university internationalisation management, as well as to assess the dynamics of its development by considering the influence of internal and external factors, which will help to effectively manage the internationalization of the university and ensure its sustainable functioning. Methods: In the paper the methodology for assessing the dynamics of the development of the international activity of a university within a time period was proposed, consisting of three main steps: reducing the dimension of the initial data, building a regression model and building an adaptive approximation predictive model. As the initial data, a system of indicators for managing the internationalization of the university at the macro, meso and micro levels was used. Results: This article presents the results of a study of the dynamics of the development of international activities of universities using multivariate component analysis, which allows us to consider many different factors, taking into account the links between them, in addition to assessing the influence of these factors on the result. A methodology has been developed for assessing the dynamics of the development of the international activity of the university within the time period, this methodology allows making a current assessment of the international activity of the university, moving into the orthogonal space of the main components and predicting further dynamics of development by constructing an adaptive approximation of the forecast model. According to the results of the forecast, universities are divided into the following categories: forcing, where international activities are actively developing; stationary, whose international activity does not change significantly and fading, universities whose international activities are deteriorating every year. Practical relevance: The research results will make it possible to implement a decision support information system for managing and evaluating the international activities of the university.

Numerous researches have established that one of the factors affecting the operational properties of mating parts is the shape error in the cross section. Modern mechanical engineering is characterized by an increasingly widespread use of numerically controlled machine tools and machining centers. A significant disadvantage of most CNC machines is that when using them, it is not guaranteed that the specified requirements for processing accuracy are obtained. The purpose of the research is to develop a system for automatic provision of non-circularity parameters during processing on a bench, including a lathe with numerical control. Research methods: Correlation analysis was used to investigate the cross-sectional profile. Carbide cutters were used. Structural steels and hard-alloy cutters were processed. The vibration signal was recorded using an accelerometer sensor. Results: It is shown that the use of a vibration signal makes it possible to evaluate the accuracy of the shape, and the use of models based on fuzzy logic and neural networks makes it possible to obtain the relationship between the parameters of the vibration signal and the accuracy of the shape in the cross section. A neuro-fuzzy system with a neural network database and an adaptive decision-making unit based on fuzzy logic is used as a device for active control (automatic provision) of shape accuracy parameters. In the course of the research, it was found that the use of such vibration signal parameters as power and correlation entropy makes it possible not only to assess tool wear, but also to implement active control of the output parameters of the processing process. The dependences of the parameters of the shape error, estimated by the coefficient of the correlation entropy and the power of the vibroacoustic signal, were obtained for different values of the wear of the cutting tool. Neuro-fuzzy models have been built to assess the state of a technological system. Practical significance: the use of neuro-fuzzy models makes it possible to determine the deviations of the shape with an error not exceeding 15%, and the use of the apparatus of fuzzy logic makes it possible to more effectively make decisions about the further use of the instrument. The results obtained were used as initial data for the development of a model of a system for automatic provision of shape accuracy parameters.