Practical issues in the detection of damage in beams using wavelets

Wavelet transform of dynamic response data, experimentally acquired using a piezoelectric sensor, is presented as a local non-destructive evaluation technique for locating damage in a beam. The higher consistency and accuracy of the results based on data from a piezoelectric sensor over a conventional strain gauge are demonstrated. The allowable range of wavelet scale to process the data is shown to be dependent on the sampling rate, filter frequency and length of signal, where edge effects due to transformation must also be considered in the latter. By estimating the wave arrival times based on theoretical flexural wave velocity, the approximate wavelet scale to process the data can be determined. Further processing of the signal at a finer wavelet scale is necessary to improve the accuracy. Based on experimental data, the method is shown to be robust with respect to boundary and damage conditions as well as impact position.