Damage Detection using Principal Component Analysis based on Wavelet Ridges

Principal Component Analysis (PCA) and Wavelet Transform (WT) are two well-known signal processing tools that are widely used in different fields. PCA plays a vital role in statistical analysis as a dimensional reduction tool. Besides, WT has proven its ability to overcome many of the limitation of the others among various time-frequency analyzers. The present work attempts to use the properties and advantages of both methodologies together in damage detection. To achieve this aim, PCA is applied on ridges of wavelet transform of measured signals from the structure. The results show that the proposed combination improves the accuracy of detection comparing with PCA damage detection based on original data captured from sensors. According to the result, when PCA uses the ridges of transformed data, the identifications of damages are more clear and accurate. This work involves experiments with an aluminum beam using piezoelectric transducers as sensors and actuators. Damages are introduced into the structure as a cut in several steps enlarging the depth of cut.