Detection of Fatigue Crack in the Shaft Using Time-Frequency Analysis

Rotating shafts operating under complex loading conditions are susceptible to develop fatigue cracks at the localized stress concentration points. Fatigue cracks are the prime sources for catastrophic failures. Therefore, it is vital and challenging for the early fatigue crack detection to prevent costly machinery breakdowns. A rotor test rig with a facility of introducing stochastic loads on to the rotating shaft is designed and developed to induce fatigue crack on the shaft is utilized for this study. Accelerated fatigue tests are performed, along by introducing stochastic loads, on a shaft specimen seeded with circumferential V-notch configuration. The methodology for early crack detection is based on analyzing the steady-state vibration data acquired, till the shaft specimen develops fatigue crack, from various locations of the test rig using different sensors like miniature accelerometers, wireless telemetry strain gauge, and laser vibrometer. In the present study, signal processing tools like continuous wavelet transforms (CWT), Hilbert–Huang transform (HHT), and short-time Fourier transform (STFT) are employed for the early fatigue crack detection. Wavelet packet (WP) analysis is used to decompose the experimental vibration response data into narrow frequency band signals before the processing of CWT, HHT, and STFT analyses. In addition, these techniques are compared to evaluate their detection performance when employed directly on raw data and preprocessed data using WP analysis. Finally, this paper provides a detailed description and comprehensive selection of time-frequency techniques for the early fatigue crack detection.