Suitability of Advanced Non-Parameterized Time-Frequency Techniques for Condition Monitoring Applications

Signals recorded in various condition monitoring applications represent important hidden signatures in time-frequency plane. In order to process highly nonstationary signals in time-frequency domain, advanced non-parameterized time-frequency representations are showing promising results. This paper investigates the applicability of non-parameterized time-frequency algorithms for condition monitoring by analyzing two distinct signal types. The signals studied encompass EEG epileptic seizures and electrical partial discharge signals. The analytical methods involve Synchrosqueezing Transform and its enhanced iterations, representing signals in their time-frequency domains with advanced sophistication. The outcomes offer a comparative assessment of various non-parameterized techniques, elucidating the preferred approach for effectively monitoring signals in such conditions.