Bayesian Approach for Main Frequency Identification on Extremely Under-Sampled Signals

This work presents Bay-Fi, a novel Bayesian approach for the direct identification of the main modal frequency from extremely under-sampled vibration signals, without requiring signal reconstruction or prior parameter tuning. The method exploits a Bayesian curve-fitting framework applied to sub-Nyquist, randomly acquired measurements, enabling robust frequency estimates in scenarios where conventional methods fail.