This study focuses on characterising the wavelet coefficients in the strain time histories of coil springs to extract bump signals that have contributed to high fatigue damage for accelerated durability analysis. Durability analysis in the time domain is commonly associated with long loading signals, which require high computational time. Through identifying high-amplitude cycles or bumps, the signal length can be reduced while still maintaining the fatigue properties. This method is commonly known as fatigue data editing (FDE). The methodology started by extracting the high wavelet energy segments in the time loading histories using the continuous wavelet transform (CWT). Analysis findings revealed that the high magnitude wavelet coefficient is highly correlated with high amplitude cycles or bumps. The efficiency of FDE to reduce the strain signal length was controlled by the gate value selected. For the resident road signal, the appropriate gate value was found to be 70% of the maximum wavelet coefficient magnitude, which gave 93.44% of signal length reduction while preserving the original fatigue damage. Meanwhile, a signal reduction of 78.23% was reported in the highway signal at a gate value of 50%. This showed that isolating the high magnitudes in the CWT wavelet coefficients can effectively extract the bump events that contributed to high fatigue damage. Hence, CWT is proven to be an appropriate technique for identifying damaging segments in random strain histories and achieving optimised signal length reduction for accelerated durability analysis.
C. et al. (Thu,) studied this question.