A Hybrid Fault Diagnosis Approach for Blade Crack Detection using Blade Tip Timing

Abstract

The harsh working environment of the compressor hinders the condition monitoring of rotating blades. Blade faults are mostly induced by foreign object damage or high cycle failure due to initial defects. Detecting these failures before blades being broken will not only ensure engine safety but largely reduce repair costs. As a non-contact measurement technique, Blade Tip Timing(BTT) becomes a more popular method recently for health monitoring and fault diagnosis. By analyzing BTT data, many blade vibration parameters could be obtained. In this paper, we discuss how to extract first bend natural frequency, amplitude and static offset from BTT data. Then the change rule of these parameters around blade fault is illustrated. Afterward, a clustering method is employed to combine these parameters. In addition, a test with two faulty blades is introduced and used to verify the effectiveness of the proposed method.