Addressing data loss in blade tip timing: A SpatioTemporal-guided Robust Compressed Sensing method

Blade Tip Timing (BTT) is a widely used non-contact technique for monitoring blade vibration in turbomachinery. However, practical applications often suffer from data loss due to sensor contamination and other issues, which limits the accuracy and robustness of traditional reconstruction methods. To overcome these challenges, this paper proposes a three-stage SpatioTemporal-guided Robust Compressed Sensing (ST-RCS) approach for BTT data reconstruction and parameter identification with data loss. First, a robust compressed sensing method is introduced to enhance resistance to outliers and non-Gaussian noise. Second, temporal stability and spatial correlation are formulated as physical priors and embedded into the compressed sensing model. Third, an iterative reweighted optimization strategy is developed to jointly estimate the signal components and adaptively adjust sample weights, which also allows for visual localization of missing data. Extensive simulations and experimental validations demonstrate that ST-RCS significantly outperforms existing methods in both reconstruction accuracy and frequency identification, particularly under high data loss rates. Tests conducted on a high-speed rotor blade test rig confirm that ST-RCS can accurately reconstruct blade tip displacement and detect both natural frequencies and harmonic components. These results highlight the method’s robustness and its potential for real-world BTT applications.