Wavelet bump extraction (WBE) for editing variable amplitude fatigue loadings
In durability testing of automobiles, load histories collected for laboratory testing are often lengthy in time. Therefore, a fatigue data editing technique is needed to summarise the load history. A fatigue mission synthesis algorithm, called Wavelet Bump Extraction (WBE) preserves the original load cycle sequences has been developed. The basis of WBE is to identify the important features or bumps that cause the majority of the fatigue damage. Bumps are identified in the frequency bands of the load spectrum using an orthogonal wavelet transform. Bumps are then extracted and combined to produce a mission signal with an equivalent fatigue damage as the onginal signal. The WBE validation was performed by analysing the cycle sequence effects in variable amplitude (VA) loadings. The experimental fatigue lives of the shortened VA loadings (Choi 2004) were compared to those predicted using strain-life fatigue damage models, i.e. Coffin-Manson, Morrow, Smith-Watson-Topper and Effective Strain Damage (ESD). The smallest difference was found between the experiment and ESD model, suggesting it is a suitable model for the use with WBE. Comparison between WBE and the time domain fatigue data editing was also conducted in order to observe its effectiveness for accelerated fatigue tests. Moreover, it is useful to evaluate the fatigue life of the original and mission signals by means of fatigue damage preservation in the mission signal. Finally, an analysis of the bump segments sequence effects was performed in order to determine an appropriate mission signal for accelerated fatigue tests. The WBE algorithm showed a substantial compression of the VA loadings could be achieved whilst maintaining fatigue damage and the important load sequences. The ability of the WBE algorithm to shorten fatigue loadings would be expected to prove useful in accelerated fatigue testing of automobiles. Finally, the combination of WBE and ESD provides a novel application of the wavelet-based fatigue data editing.