Розробка спеціального програмного математичного забезпечення управління вентиляційними установками нового безпечного конфайнменту ЧАЕС

• Дата: 2022
• Автори: Pysmennyy, Y., Havrylko, Y., Krukovskyi, P., Starovit, I., Diadiushko, Y.
• Формат: Стаття
• Мова: Ukrainian
• Опубліковано: State Scientific and Technical Center for Nuclear and Radiation Safety 2022
• Онлайн доступ: https://nuclear-journal.com/index.php/journal/article/view/949
• Назва журналу: Nuclear and Radiation Safety

The processes of air masses movement with radioactive aerosols (RA) both inside and outside the Shelter’s New Safe Confinement (NSC) of the Chornobyl Nuclear Power Plant are studied. The NSC is a relatively airtight structure but, because of larges sizes, has leaks at the joints of structures, the total area of which can reach 15 m2. The process of air circulation and leakage is significantly influenced by hydrometeorological conditions, especially wind direction and speed, temperature, and humidity. The processes of RA mass transfer are described by a complex hydraulic model of the facility, which depends on both external conditions and the operation mode of internal equipment. The equipment that provides control of the air masses movement in the NSC includes ventilation systems (VS). Their main purpose is to control the movement of large volumes of air inside and to avoid and minimize unfiltered RA emissions outside the NSC. Analysis of the information received from the institution operating the NSC allowed the conclusion that the decision-making process for optimal management of ventilation equipment was possible only with continuous processing of various data that change throughout the NSC and with constant changes made to the parameters used, which is a complex technical problem. With the aim of solving such a complex technical problem, it is necessary to solve a scientific problem, which leads to the development of mathematical software based on machine learning methods, especially neural networks, which would be able to take into account multifactor indicators and find the optimal control algorithms. The aim of the work is to optimize the operation of VS to ensure minimum RA emissions outside the NSC with restrictions on the cost of electricity in real time. The article, in accordance with the analyzed factors influencing uncontrolled air emissions with radioactive aerosols, proposes mathematical software. Neural network training was conducted on the basis of actual indicators of air circulation inside and outside the NSC, which allowed the problems to be solved and the research goal to be achieved.