WAVELET BUMP EXTRACTION (WBE) ALGORITHM FOR THE ANALYSIS OF FATIGUE DAMAGE

This paper presents a new algorithm known as Wavelet Bump Extraction (WBE) for summarising long records of fatigue data by extracting the important features that cause the majority of the damage. One of the main elements in WBE is the orthogonal wavelet transform. In WBE the input signal is decomposed using Daubechies wavelets and the wavelet levels are grouped into characteristic frequency bands. Features, called bumps, are extracted from each wavelet group and they are later synchronised together to produce a bump signal. The accuracy of the WBE algorithm has been evaluated by application to two experimentally measured data sets. One data set had a predominantly tensile mean load and the length of the extracted bumps was 41 % of the signal. The fatigue damage ratio between the original and reduced time histories was 96 % and 99 % as calculated using the nSoft® package with the Smith-Watson-Topper and Morrow strain-life models, respectively. The other data set had predominantly compressive mean load and the bump signal length was 55 % of the original signal. The fatigue damage ratio between the original and reduced time histories was 97 % and 90 % as calculated using the Smith-Watson-Topper and Morrow strain-life models, respectively. The results suggest that the WBE algorithm is an efficient approach for summarising long records of fatigue data