A novel parameter estimation method of BTT data without gridding error

As a critical component of rotating machinery, the condition monitoring of blades is of significant value. Blade Tip Timing (BTT), a widely used non-contact measurement technique, effectively captures blade vibration information through a specially arranged set of probes. While extracting vibration parameters from BTT signals is no longer a challenge, enhancing estimation accuracy and mitigating the impact of gridding errors remains crucial. In this paper, an off-grid iterative algorithm based on sparse Bayesian inference is proposed. By hypothesizing the various parameters of the BTT signal receiving model, the corresponding joint probability density function is obtained. Further, the Bayesian theorem is employed to infer the posterior probability distribution, and the Expectation-Maximization algorithm is used to iteratively update the frequency error weights, signal power, and noise variance. This approach effectively eliminates the influence of gridding errors and achieves high-precision parameter estimation. The proposed method has been validated using data from both simulation and a real test rig, demonstrating its superior performance over existing methods. The precision improvement from "off-grid"enables more reliable and earlier detection of changes in blade vibration parameters.