Rotor Blade Fault Detection through Statistical Analysis of Stationary Component Vibration

摘要

Rotor blade fault detection and health monitoring systems are crucial for gas turbine engine testing and evaluation. The most commonly used techniques involve drilling optical access holes in the engine casing for probes to directly monitor blade deflection and vibration. In this work, a less intrusive, indirect technique for rotor blade fault detection is developed. The method utilizes the signal from an accelerometer attached to a stationary, stator-like probe located just downstream of the rotor system. The vibration of the stator was processed to determine the relative locations of the vibratory peaks, which are assumed to be directly related to the relative locations of the rotor blades. The current study shows that the resulting stator vibrations indeed contain rotor blade location information, similar to the blade tip timing information gathered from a non-contact stress measurement system. Statistical analysis of the vibratory peak location data was used to identify one type of seeded blade fault (an intentionally offset blade). The seeded offset amplitude and the seeded offset blade rotor position were varied to determine the effectiveness of the fault detection algorithm. The results show that the fault detection technique is able to detect the presence, rotor position, and change in offset amplitude of the offset blades.