A novel self-adaptive, multi-peak detection algorithm for blade tip clearance measurement based on a capacitive probe

Abstract

Peak height detection is critical for blade tip clearance (BTC) measurement based on capacitive probes. However, the noise interference of the complicated working environments and instabilities of the rotation speed make high-precision peak detection a challenge. To solve this problem, a novel self-adaptive, multi-peak detection algorithm is proposed in this paper. The adaptive dual thresholds are used to identify the single peak and pre-estimate its peak parameters for the segmentation of the peak support region and the selection basis of the wavelet parameters. Then, based on the prior pre-estimation information, the wavelet matching is performed on the segmented single peak adaptively for eight times, and the derived mathematical models of the wavelet matching are fitted using the obtained matching coefficients to get precise peak parameters such as peak height and peak position. The blade tip clearance is calculated from the peak height through the calibration curve. The experiments show that the BTC measurement error of the proposed algorithm is under 15 μm, with the clearance ranging from to , and the proposed algorithm has preferable stability to comparable algorithms with the rotation speed ranging from 1000 rpm to 4000 rpm. This implies that the proposed algorithm has high anti-noise performance and good adaptability to rotation speed change.