Russian Federation
An approach to ensuring IT infrastructure fault tolerance based on noise data analysis and machine learning is considered. Sources of "noise" in IT systems are analyzed: logs, metrics, and acoustic signals. Architectures of models detecting anomalies 6-10 days before failure are studied. Problems of interpretability and data quality are considered. Promising directions for predictive analytics development are analyzed.
IT infrastructure, predictive analytics, machine learning, data mining, anomaly detection, AIOps, observability
1. Zhigalov K.Yu., Ogorodnikov O.V. Robastnaya sistema monitoringa tehnicheskogo sostoyaniya stroitel'nyh i dorozhnyh mashin pri nepolnyh i zashumlennyh dannyh sensorov // Stroitel'nye i dorozhnye mashiny. 2025. №10, Tom 69. S. 102-108. URL: https://www.ipu.ru/node/82773 (data obrascheniya: 19.02.2026).
2. Rossiyskaya sistema monitoringa Datcheck poluchila II-moduli dlya prognozirovaniya sboev // TAdviser. 2025. URL: https://www.tadviser.ru/index.php/Produkt:Datark_Datcheck_SCADA-sistema (data obrascheniya: 19.02.2026).
3. Remil Y. A data mining perspective on explainable AIOps with applications to software maintenance: Ph.D. thesis. INSA de Lyon, 2023. URL: https://theses.hal.science/tel-04391281 (data obrascheniya: 19.02.2026).
4. Neuron Soundware and CEITEC BUT bring intelligent machine lifespan prediction // RICAIP. 2025. URL: https://ricaip.eu/neuron-soundware-machine-lifespan-prediction/ (data obrascheniya: 19.02.2026).
5. Zhang L. Obnaruzhenie anomaliy ChPU s pomosch'yu mashinnogo obucheniya bez uchitelya. Chast' 4: Primenenie na proizvodstve // isicad.ru. URL: https://isicad.ru/ru/articles.php?article_num=21229 (data obrascheniya: 19.02.2026).
6. Rossiyskaya sistema monitoringa Datcheck poluchila II-moduli dlya prognozirovaniya sboev // CNews.ru. 2025. URL: https://www.cnews.ru/news/line/2025-12-05_rossijskaya_sistema_monitoringa (data obrascheniya: 19.02.2026).
7. Remil Y. A data mining perspective on explainable AIOps with applications to software maintenance // HAL Université Lyon 2. 2023. URL: https://hal.univ-lyon2.fr/tel-04391281v1/datacite (data obrascheniya: 19.02.2026).
8. Audible Noise-Based Hardware System for Acoustic Monitoring in Wind Turbines // MDPI Inventions. 2025. Vol. 10, № 4. Art. 58. URL: https://www.mdpi.com/2411-5134/10/4/58 (data obrascheniya: 19.02.2026). DOI: https://doi.org/10.3390/inventions10040058
9. Sistema raspoznavaniya nepoladok na proizvodstve na osnove akusticheskogo kanala s pomosch'yu iskusstvennogo intellekta // Universitet 2035. URL: https://pt.2035.university/project/sistema-raspoznavania-nepoladok-na-proizvodstve-na-osnove-akusticeskogo-kanala-s-pomosu-iskusstvennogo-intellekta (data obrascheniya: 19.02.2026).
10. Observability in 2025: From Reactive to Predictive // ETR.ai. 2025. URL: https://research.etr.ai/etr-data-drop/observability-in-2025-from-reactive-to-predictive (data obrascheniya: 19.02.2026).
