Subspace Dimension Reduction for Faster Multiple Signal Classification in Blade Tip Timing

Abstract

Blade tip timing (BTT) is a noncontacting vibration measurement method for rotating blade condition monitoring. The BTT sampling frequency is proportional to the number of probes under uniform probe layout and constant rotating speed. However, the limitation of the probe installation space in aeroengines causes the lack of probes, which leads to the inherent undersampled feature of the BTT signal. Thus, frequency identification of nonuniform undersampled signal has become a popular field in BTT signal processing. Multiple signal classification (MUSIC) is an array signal processing method that has been applied to this field. Nevertheless, the frequency traversal in MUSIC is a time-consuming process that will affect the online monitoring performance. Its running time is highly related to the dimension of the noise subspace. The subspace dimension reduced MUSIC (SDR-MUSIC) is proposed in this article to reduce the running time of MUSIC by reducing the dimension of noise subspace. Simulations and experiments are conducted to testify that SDR-MUSIC can significantly reduce the running time of MUSIC while ensuring accuracy.