UDC 004.89
UDC 614.842.83
The article discusses the forecasting of fires in the Republic of Abkhazia using modern methods of data mining and machine learning. The article begins with an analysis of statistical data on fires in the Republic from 2013 to 2022. The graphs show data on the frequency and distribution of fires by time and territory, as well as material damage caused by fires in each of the regions of the Republic. Data on the number of preventive measures carried out and the number of deaths from fires are presented. The analysis makes it possible to identify key patterns and trends that will serve as the basis for the development of a predictive model, which, in turn, will improve the effectiveness of fire safety management. The next part of the article describes the initial data of the task of developing a neural network in Python, expressed in requirements and constraints. The architecture of the model includes an input layer, hidden layers, and an output layer responsible for predicting the number of fires and deaths in the future. Python libraries such as xlrd, pandas, numpy, neurolab, Matplotlib, and tensorflow were used to implement the neural network. The stages of model training, validation and testing of the obtained results are considered. The final part of the article provides a comparison of the results of the developed neural network with the results of networks available on the Internet, such as deepseek and ChatGPT. The strengths and weaknesses of various models are shown, as well as potential methods for improving predictive abilities. In conclusion, it is pointed out the importance of further research in the field of integrating various data sources and machine learning methods to improve the accuracy of forecasts and the effectiveness of fire safety management
prognozirovanie pozharov i gibeli lyudey, neyronnaya set', statisticheskiy analiz, upravlenie pozharnoy bezopasnost'yu
1. Zakon O pozharnoy bezopasnosti: Zakon Respubliki Abhaziya ot 21.07.2021 No 5180-s-VI // Prezident Abhazii: sayt. - URL: http://presidentofabkhazia.org/upload/iblock/af1/ZRA-O-POZHARNOY-BEZOPASNOSTI.pdf.
2. Zaharov V. Intellektual'nye tehnologii v sovremennyh sistemah upravleniya // Problemy teorii i praktiki upravleniya. - 2005. - No 4. - S. 96-100. EDN: https://elibrary.ru/KVSPMD
3. Pranov B.M. O nekotoryh podhodah k modelirovaniyu i prognozirovaniyu vremennyh ryadov pozharnoy statistiki // Tehnologii tehnosfernoy bezopasnosti. - 2014. - No 5 (57). - S. 5. - URL: http://ipb.mos.ru/ttb. EDN: https://elibrary.ru/TLPNLZ
4. Kaybichev I.A., Kaybicheva E.I. K voprosu ob adekvatnosti modeli Kobba - Duglasa pri prognozirovanii vremennyh ryadov pozharnoy statistiki // Tehnosfernaya bezopasnost'. - 2019. - No 2 (23). - S. 3-15. EDN: https://elibrary.ru/KSCWOI
5. Kaybichev I.A., Kaybicheva E.I. Regressionnyy analiz vremennogo ryada kolichestva pozharov v Rossii // Sibirskiy pozharno-spasatel'nyy vestnik. - No 2. - S. 49-53.
6. Baturo A.N. Prognozirovanie kolichestva pozharov v regione na osnove teorii vremennyh ryadov // Tehnologii grazhdanskoy bezopasnosti. - 2013. - Tom 10, No 3 (37). - S. 84-88. EDN: https://elibrary.ru/REMCHR
7. Materov E.N. Ispol'zovanie yazyka programmirovaniya R v voprosah pozharnoy bezopasnosti: analiz statistiki kolichestva pozharov na osnove teorii vremennyh ryadov // Sibirskiy pozharno-spasatel'nyy vestnik. - 2019. - No 1. - S. 52-57. EDN: https://elibrary.ru/PAZTYM
8. Materov E.N. Ispol'zovanie yazyka programmirovaniya R v voprosah pozharnoy bezopasnosti: analiz glavnyh komponent // Sibirskiy pozharno-spasatel'nyy vestnik. - 2019. - No 2. - S. 49-53. EDN: https://elibrary.ru/TZYISD
9. Nacional'naya associaciya protivopozharnoy zaschity (NFPA). Uscherb ot pozharov v SShA. - 2020. - URL: https://www.nfpa.org/.
10. Lyu Yu., Chzhan M. Prognozirovanie riska vozniknoveniya pozhara s pomosch'yu mashinnogo obucheniya: obzor // Modelirovanie okruzhayuschey sredy i programmnoe obespechenie. - 2021. - Tom 137. - S. 104977. -. DOI:https://doi.org/10.1016/j.envsoft.2021.104977
11. Li Chzhunchzhi. Prognozirovanie ploschadi, ohvachennoy lesnymi pozharami, s pomosch'yu neyronnoy seti s dolgovremennoy kratkovremennoy pamyat'yu i mehanizmom vnimaniya // Pozharnaya tehnika. - 2021. - Tom 57. - S. 1-23. -. DOI:https://doi.org/10.1007/s10694-021-01142-4
12. Oleynikov V.T., Mosyagin A.A. Vozmozhnost' prognozirovaniya opasnyh situaciy v sub'ektah RF na osnove neyronnyh setey // Institut problem bezopasnosti (IPB): sayt. - URL: http://www.ipb.mos.ru/konf/2004/sb-2004/sec-2-04/2.68.pdf.
13. Stankevich T.S. Razrabotka metoda operativnogo prognozirovaniya dinamiki razvitiya lesnogo pozhara s pomosch'yu iskusstvennogo intellekta i glubokogo mashinnogo obucheniya // Vestnik Irkutskogo gosudarstvennogo tehnicheskogo universiteta. - 2018. - T. 22, No 9. - S. 111-120. EDN: https://elibrary.ru/YGGSLR
14. Haykin S. Neyronnye seti: polnyy kurs: per. s angl. / pod red. N.N. Kussulya, A.Yu. Shelestova. - 2-e izd., ispr. - M.: Izdatel'skiy dom "Vil'yams", 2008. - 1103 s.
15. Iskusstvennye neyronnye seti i prilozheniya: ucheb. posobie / F.M. Gafarov, A.F. Galimyanov. - Kazan': Izd-vo Kazan. un-ta, 2018. - 121 s.
16. Neyronnaya set' Deepseek [Elektronnyy resurs] // Deepseek Chat: sayt. - URL: https://chat.deepseek.com/.
17. Neyronnaya set' ChatGPT [Elektronnyy resurs] // OpenAI: sayt. - URL: https://chatgpt.com/.
18. Dmitrieva E.G. Problemy zaschity personal'nyh dannyh v cifrovom mire i puti ih resheniya // Pravo i biznes. - 2021. - No 3. EDN: https://elibrary.ru/MQYBJB
19. Bychkov A.I. Problemy zaschity personal'nyh dannyh. - M.: Infotropik Media, 2020. - 116 s.
