Rub-impact fault diagnosis for rotating blade using covariance-reconstructed MVDR and BTT data

Abstract

Blade Tip Timing (BTT) is a crucial technique for monitoring the health status of rotating blades in aero-engines. However, the inherently severe non-uniform under-sampling characteristics of BTT signals make the extraction of early fault features highly challenging. To address the aforementioned issues, this paper introduces the principle of Minimum Variance Distortionless Response (MVDR) method into the problem of blade fault feature identification, and proposes a MVDR algorithm based on Covariance Reconstructed to achieve rub-impact fault diagnosis of rotating blades. By utilizing subspace orthogonality to reconstruct the covariance matrix, this method effectively suppresses the spurious peak interference generated by the traditional MVDR under limited snapshot conditions, while reducing the practical computational burden under limited snapshot conditions. The results of numerical simulations and rotating blade rub-impact tests indicate that the proposed method can accurately recover the fractional harmonic features corresponding to rub-impact faults from severely under-sampling BTT data, including the n X/2 and n X/3 harmonic series, and effectively identify the nonlinear vibration responses under different rub-impact states. The relevant identification results maintain excellent consistency with the analysis results of the synchronously acquired strain signals. The research findings demonstrate that this method can effectively achieve feature extraction and early warning of rotating blade rub-impact faults, providing a feasible technical approach for the online health monitoring of aero-engine blades.