Wavelet based algorithms for crack detection in rotor-machinery

Initiation and propagation of surface cracks in rotary shafts are known causes for dynamic failure modes that could significantly impact the reliability and operation safety of high-speed rotor-dynamical machines. Early detection of the on-set and progression of crack-induced system instability and chaotic response would considerably extend the lifetime and improve the reliability of the mechanical system. A wavelet-based algorithm effective in identifying the initiation and progression of mechanical instability including chaotic response has been applied to determine the nonlinear dynamical characteristics of a cracked rotor model. The algorithm effectively correlates the breathing action of surface cracks with rotor dynamic motions in the simultaneous time-scale wavelet domain. Rotor-dynamic instabilities representing period-doubling, fractal-like, and chaotic system responses are considered to demonstrate the effectiveness of the algorithm in detecting mechanical faults as contrast to algorithms that are based on nonlinear dynamics. The results not only just show the feasibility of the algorithm for mechanical fault diagnostic but also suggest its applicability to in-line, real-time condition monitoring at both the system and component levels.

Wavelet based algorithms for crack detection in rotor-machinery Academic Article