Wavelet-based Feature Extraction Algorithm for Fatigue Strain Data Associated with the k-Means Clustering Technique

The study presents the development of a wavelet-based segmentation algorithm for fatigue life assessment. Strain data was extracted using the Morlet family. The extraction process identified damaging segments, and it was able to shorten the original signal by 74.3%, with less than 10% difference with statistical parameters. The extraction algorithm was able to retain at least 97.9% of fatigue damage. The damaging segments drawn were clustered using the k-means method to provide three groups of segments, i.e., lower, moderate, and higher groups representing statistical values. The approach was suggested as an alternative method for evaluating and clustering fatigue strain signals.