Forgot your password?
scientific activity
structureeducational projectsperiodicalsstaffpress centercontacts
­ˇ˝˝ŕŔÚ | english

Structural Models and Deep Learning

Joint IITP RAS, NRU HSE &Skoltech seminar  

November 15 (Tuesday), 1830, IITP, room 615    
Evgeny Burnaev (joint work with Pavel Erofeev, Ivan Nazarov, Dmitry Smolyakov)
Anomaly-to-failure Matching for Predictive Maintenance of Complex Systems
Predictive analytics and modeling become more and more important for maintenance of complex technical systems. Introduction of condition-based maintenance strategies in different industries significantly reduces overall maintenance costs and improves reliability. Modern complex technical systems, e.g. aircraft engines, are fully equipped with batches of sensors and detectors allowing for real time monitoring and diagnostics. Typically equipment falls into pre-failure state starting with some minor flaws, e.g. cracks or leaks, that evolve in time and lead up to critical failure events such as complete engine destruction. Natural need of the maintenance engineers is to be able to identify this flaws as early as possible and thus try to prevent or even avoid critical events. Or at least be able to prepare for the event on time. In some cases based on real time sensor observations it is possible to indirectly identify anomalies in system behavior related to the minor problems. That is where development of accurate and reliable mathematical models and tools comes up to the stage. In some industries, e.g. aviation, it is crucial to have models with maximum predictive power and strictly limited false alarm rate. In this presentation we specify generic validation framework for failure prediction models and propose a novel exhaustive methodology for anomaly-to-failure matching. We prove effectiveness and reliability of the proposed methodology in the specified validation framework on the real test case from aerospace industry. 
13.11.2016 | Efimova Maria


  © Institute for Information Transmission Problems of the Russian Academy of Sciences (Kharkevich Institute), 2017
About  |  Contacts  |  The old version of this site