Dual-Parameter Diagnosis of Gas Turbine Blade Fatigue Cracks Based on Blade Tip Timing Data

Gas turbine rotor blade cracks are usually caused by high-cycle fatigue. The natural frequency and damping of the blades change as the cracks propagate. Most existing methods detect fatigue cracks by observing the decrease in natural frequency, which may lead to misjudgment. This article proposes a method for diagnosing blade fatigue cracks by combining the indicators of natural frequency decrease and damping ratio increase. Extracting these two parameters using blade tip timing (BTT) data faces challenges such as signal undersampling and errors in identifying the resonance location due to different sensor layouts. This article introduces a method for extracting natural frequency and damping using only two sensors under uniform acceleration (sweep frequency) excitation. For natural frequency extraction, an adaptive window length sparse representation method is proposed, which can adaptively select the spectral window parameters based on the sampled data and effectively reduce leakage by adding weighting factors. For damping extraction, a center-fixed half-power bandwidth method is proposed. This method improves the half-power bandwidth method by locating the resonance center based on the extracted natural frequency, avoiding interference from “dip” data caused by different sensor layouts. Finally, an accelerated fatigue experiment on a blisk verifies the accuracy of the extracted parameters and the sensitivity of the natural frequency and damping indicators to fatigue cracks.