Abstract
Blade tip timing (BTT) is a non-contact measurement method for rotor blades. Its non-uniform sampling pattern is determined by the physical probe placement and rotational speed. Due to the lack of sampling probes and uneven placement, the anti-aliasing spectrum analysis for non-uniformly sampled signals becomes a hotspot in the BTT field. In this paper, a forward backward spatial smoothing (FBSS) method is used to estimate the autocorrelation matrix more accurately, which enables a better frequency identification ability for the autocorrelation matrix-based methods. Additionally, the relationship between the exponential complex steering vector and real-valued steering vectors is revealed. The peak significance is proposed to measure the quality of the pseudo spectrum. By taking multiple signal classification and minimum variance distortionless response as two example methods, the superiority of FBSS is demonstrated by simulations and experiments.
