Статистический анализ и моделирование данных о ликвидации пожаров на топливно-энергетических предприятиях

Abstract

PURPOSE. Using the example of statistical data on fires at fuel and energy enterprises for the period from 2000 to 2020, stored in the federal database “Fires”, an algorithm for these data statistical analysis is proposed for constructing a statistical digital twin (SDT) of the type of fires under consideration. The SDT under development is aimed at creating any necessary and adequate to the initial data sample set of records (dataset) for the simulated class of fires for subsequent modeling and other studies. The objectives of this work are the construction of an algorithm and the study of the main key points in constructing the SDT. The object of the study is a set of fires, whose characteristics are enclosed contained in the form of separate records in the database, while the subject of the study is a set of statistical static and dynamic models that should be built on the basis of these data analysis. METHODS. Methods of probability theory, mathematical statistics, simulation modeling and statistical estimation are used. An approach based on the joint use of the Monte Carlo method and bootstrap principles to form adequate training samples is proposed. FINDINGS. MCB-modeling algorithm (Monte Carlo bootstrap) is proposed for generating a statistical sample of data on fires in the amount necessary for machine learning of a wide class of decision support models. The relevance of the proposed approach is due to the fact that in some cases the available statistics are insufficient in volume for building decision support models and other studies. RESEARCH APPLICATION FIELD. SDT, built on the basis of MCB-modeling, will allow forming data samples of the required volume, which are similar to the original databases. CONCLUSIONS. The application of the proposed algorithm for constructing SDT for MCB-modeling will allow building machine-learning models for their use as part of decision support systems in safety systems.