Oleg Ilyunin
Verified @nure.ua
Kharkiv National University of Radioelectronics / Department of Computer Intelligent Technologies and Systems
EDUCATION
Candidate of Technical Sciences, Senior Lecturer of the Department of Computer Intelligent Technologies and Systems, Kharkiv National University of Radioelectronics, Kharkiv, Ukraine
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Science, Artificial Intelligence, Computer Science Applications, Computer Vision and Pattern Recognition
FUTURE PROJECTS
Study of Recurrent Neural Models predicting Deposits in Plate Heat Exchangers
The efficiency of plate heat exchangers (PHEs) included in process production lines depends on the cleanliness of the plate surface and directly affects the quality of the final product, fuel consumption of heat generating plants and carbon emissions. Scheduling routine maintenance for PHE cleaning increases the efficiency of heat exchange systems operation. Until recently, complex mathematical modeling was used to predict the value of the heat transfer coefficient after a certain period of operation of the heat exchanger and the point in time when the coefficient reaches the allowable limit, using systems of differential equations and matrices of heuristic coefficients, which required serious computing resources. This paper presents a study of the performance of neural network models for solving this problem: a standard recurrent neural network (RNN) with fading gradient and RNN with two hidden layers LSTM (long short-term memory) which can learn long-term dependencies, process sequen
Applications Invited
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
Oleg Ilyunin, Oleg Rudenko, Oleksandr Bezsonov, Stanislav Boldyryev, Viktor Zorenko, and Nataliia Serdiuk
IEEE
The article is devoted to the issue of constructing a fuzzy classifier, with the help of which the approximation of real data sets is carried out for the construction of a neuroregulator for the controlled temperature of working solutions of the technological process of pickling rolled metal. The relevance of the topic is to improve the energy efficiency of technological pickling for the metallurgical industry. The development of a technique for constructing a neuroregulator for the temperature of a pickling solution with a given control accuracy is the main task of the work. The results of modeling the characteristics and approbation of the neuroregulator showed a reduction in the consumption of thermal energy by the technological process by 8%. With appropriate adaptation of the model, it is possible to use the neuroregulator in the field of public utilities, at chemical enterprises.
Oleksandr Bezsonov, Oleg Ilyunin, Botagoz Kaldybaeva, Oleksandr Selyakov, Oleksandr Perevertaylenko, Alisher Khusanov, Oleg Rudenko, Serhiy Udovenko, Anatolij Shamraev, and Viktor Zorenko
SDEWES Centre
Steel pickling processes are very important for steelmaking production quality. Pickling process is based on chemical reaction of acidic pickling solution with scale impurities on steel strip surface. In sulfuric acid pickling process together with scale removal. The partial dissolving of steel surface takes place because of sulfuric acid attack takes place. * Corresponding author Bezsonov, O., et al. Resource and Energy Saving Neural Network-Based ... Year 2019 Volume 7, Issue 2, pp 275-292 Journal of Sustainable Development of Energy, Water and Environment Systems 276 Continuous sulfuric acid carbon steel pickling in existing plants is very energy and water consumptive. An innovative approach is proposed for modernization of continuous sulfuric acid pickling process performance. The proposed neural network model may be used to optimize consumption of sulfuric acid, decrease energy consumption, reduce steel losses and, respectively, reduce harmful wastes and emissions from continuous steel pickling lines. This is possible because of quick adaptation of neural network model to changing environment through fast training algorithms. The developed model identifies the temperature necessary to provide the set process rate at the current variable values of the parameters: concentration of sulfuric acid and concentration of ferrous sulfate multi-hydrates in solution and transmits the temperature value as a current task to regulator in each discrete moment of the process. The results of application of the developed neural network, included as a part of the presented process supervisor, prove its efficiency in use for pickling process operational control: steam consumption for pickling process was decreased by 8%, acid consumption for pickling process was decreased by 26%, while the process efficiency and quality remain unaffected.
Anatoliy Shamraev, Elena Shamraeva, Anatoly Dovbnya, Andriy Kovalenko, and Oleg Ilyunin
Springer International Publishing
Stanislav Boldyryev, A. O. Garev, O. Ilunin, A. Shamraev, O. Selyakov, O. Leshchenko and P. Kapustenko
In this work the cost-effectiveness of coal mine waste water low potential heat utilization and integration in the enterprise local heating network by the bivalent parallel scheme is investigated. It is shown that the additive criterion of economic efficiency is not always sensitive to the target value because of the reason of non-linear coupling between the arguments of the target function. Several cases were considered from the viewpoint of cost-effectiveness criterion. Moreover, the arguments of the target function are also nonlinear, depending on technological parameters of the heat pump (HP) equipment. As a more effective criteria were proposed the modified canonical additive-multiplicative function and Kolmogorov-Gabor polynomial function to obtain the generalized multivariate estimation of alternatives in two stages procedure.
RECENT SCHOLAR PUBLICATIONS
MOST CITED SCHOLAR PUBLICATIONS
RESEARCH OUTPUTS (PATENTS, SOFTWARE, PUBLICATIONS, PRODUCTS)
Device for continuous pickling of rolled carbon steel
Ukrainian Patents Database |
2014-02-25 | Patent
PAT: 104710 Ukraine
URL