Utilization of advanced signal procession methods in detection of early stages of machine failures

Aim of project
The aim of the project is elaboration of methodology of detection of early stages of failure development, which is the basis of a diagnostic system integrated with the object. Such a system allows for automatic reduction of probability of catastrophic failure occurrence.

During the project, tasks connected with adaptation, elaboration and verification of methods of analysis, separation, classification and diagnostic-forecasting inference are implemented. Research done within the project focuses on detection of causes of failures, their low-energetic development stages. Detection is made using analysis of non-liner effects and modulation effects accompanying occurrence of failures.

Research on modeling algorithms of amplitude-frequency modulation phenomena and effective demodulation methods, cyclestationary analysis, use of wavelet analysis, time-frequency representations and higher order spectra analysis is conducted.

Special attention is paid to research of use of linear algorithms for analysis of non-linear diagnostic relationships describing appearance phenomena and development process of particular failure stages. Efficiency and effectiveness assessment of the elaborated methods will be carried out on the basis of results received from laboratory test stands.

Expected results
Adaptation and elaboration of new methods of signal analysis for detection of early stages of machine failures;

Utilization of artificial intelligence, blind separation method and blind filtration in the tasks of separation and classification of early failure stages;

Development of diagnostic methods based on artificial intelligence;

Experimental verification of the elaborated methods on laboratory test stands;

Publication of results.