Non-contact crack detection of high-speed blades based on principal component analysis and Euclidian angles using optical-fiber sensors

High-speed rotating blades are key mechanical components in turbo-machinery. High cycle fatigues often induce blade cracks and ultimately the loss of blades. Effort has being made to measure on-line blade vibrations and promptly detect potential cracks. Traditional contacting methods are difficult to satisfy the needs of on-line testing, enduring extreme conditions and non-invasion. Nowadays non-contact vibration measurement at blade tips by optical sensing has become a promising approach. In this paper a blade tip-timing (BTT) method using four optical-fiber sensors is presented and all-blade vibration displacements are derived in detail. Blade vibration signals collected by the BTT method are typical sub-sampled signals, so a signal reconstruction method is proposed based on the Shannon sampling theorem. Next nine candidate vibration parameters are calculated to form the crack feature space and principal component analysis (PCA) is used to extract principal components from it. Then Euclidian angles of principal components are defined to detect cracks. In the end, an experimental set-up is built and a small crack is cut on a blade artificially. Collected data are analyzed to testify the proposed method and the experimental results demonstrate its effectiveness.