EU program "Horizon"
"Horizon 2020" ("Horizon 2020") is a program of the European Union that was launched in 2014 and is intended to finance research and innovation in Europe. It is the largest research and development funding program in the world, with a budget of more than 80 billion euros for the period from 2014 to 2020. The program aims to support innovative projects in such fields as science, technology, health, energy, transport, environment, social sciences and humanitarian studies. Participation in the program is open to all countries, which involves scientists, entrepreneurs, governments, organizations and public organizations from different countries of the world.
In 2021, a new program cycle started - "Horizon Europe" ("Horizon Europe"), which is the successor of Horizon 2020 and covers the period from 2021 to 2027.
However, a number of projects funded by the Horizon 2020 program are still being implemented. In particular, INNOVINNPROM LLC won "Asset Performance Management System for grain processing industry SAKURA-APM, PaaS SAKURA-IIoT based» with financial support from the research and innovation program of the European Union "Horizon 2020" within the framework of the project BOWIE, which is financed under grant agreement No. 873155, in which the NDL of the VNTU department is the official subcontractor for the performance of works for 2022-2023. In particular, by scientists of the National Institute of Artificial Intelligence and Cognitive Science (SHIK)and NDL EDEM of the Department of SAIT, the National Development Program No. 2860 "Development of information technologies to optimize grain elevator operation using neural network models and reinforcement learning methods" is being carried out (research supervisor - Mokin V.B., the responsible executor is a teacher of the faculty. SITE Dratovanyi M.V.) by order of INNOVINNPROM LLC with EU funding. Taking into account the international financing of the project, the reporting of the GDR is conducted in both Ukrainian and English languages.
Teachers, engineers and graduate students and students of the SAIT department of specialties 124 System analysis and 126 Information systems and technologies participate in the implementation of the project for a fee. Joint seminars of project teams with VNTU and INNOVINNPROM LLC both online in Meet, Zoom, Telegram, Viber, and offline. In particular, an offline seminar of the National Technical University of Artificial Intelligence and Cognitive Science was held to discuss the optimization of the architecture and parameters of neural network models of grain elevator nodes based on real data. And on January 18, 2023, a joint seminar on the topic "Development of information technologies to optimize grain elevator operation using neural network models and reinforcement learning methods" was held on the basis of NDL SHIK.
As a result GDR No. 2860 VNTU is a subsystem for optimizing grain elevator energy saving using artificial intelligence technologies to process information collected by the system "SAKURA-ARM" INNOVINNPROM LLC based on Internet of Things technologies. In particular, VNTU has developed an information technology for building a bank of neural network models of grain elevator nodes pretrained on retrospective data, which in the future allow forecasting the total energy consumption and average load of given elevator routes and their variations. And the subsystem created at VNTU allows you to choose the optimal route (variation) with the help of this bank of pre-trained models in one of two ways, depending on the number of permissible variations of the route at a given moment: for a small number of variations - forecasting and direct selection of all permissible variations, for a large number - selection optimal option using machine learning with reinforcement. The subsystem is integrated into the SAKURA-ARM system of INNOVINNPROM LLC and displays results in the form of dashboards and information panels of the system itself. If the operator chooses a sub-optimal option of the route or variation, the system will show him the best one with predictive parameters from the subsystem developed at VNTU. In addition, it is periodically analyzed whether forecasts have been confirmed regarding the initial parameters of nodes of the selected route and the route as a whole. Based on the results of the analysis using a certain algorithm, a conclusion can be drawn either about the need to improve the bank of pre-trained models, or about the failure of certain elevator devices. The subsystem was tested on the basis of real minute-by-minute observations collected over several years at one of the elevators in Ukraine. The constructed neural network models provide a relative accuracy of predicting the output parameters of the nodes up to 10%, which is quite high for such noisy data.
The system can be adapted to other similar productions with a multi-stage technological cycle. There are already a number of orders, as high-tech energy-saving systems of the "Industry 4.0" level are becoming more and more relevant.
Some of the results of the project are used in the educational process for teaching disciplines: "Internet of things and intelligent data analysis", "Information technologies of monitoring and data analysis", "Artificial intelligence technologies", "Intelligent information technologies", "Management of complex systems", "Optimization of management dynamic objects", "Information technologies for monitoring and analyzing the state of complex systems", etc.
The system was created by a joint team of "INNOVINNPROM" LLC and the department of SAIT VNTU (NDL SHIK and NDL EDEM). The process of work on the project has been released video clip.
The completed set of works made it possible to find new interesting tasks for scientific research, which will be solved in dissertations for obtaining a doctor of philosophy in the specialty 124 Dratovanyi M.V. (supervisor - Prof. Mokin V.B.) and others, as well as in numerous master's and bachelor's qualification theses of students majoring in 124 System Analysis and 126 Information Systems and Technologies.