A Study on Validation of Fatigue Damage Clustering Analysis Technique Based on Clustering Validation Index

This paper presents the comparative study on two types of the clustering technique for decomposing Variable Amplitude (VA) loadings signals based on its amplitude. These two techniques are used to recognize clusters or patterns of fatigue damaging events in the record which will bring aboutthe majority of fatigue damage. However, one of the problems that existswhencomparing which technique will produce better clusters is the fact thata clustering validation index isneeded. In this study, techniques that were used were theFuzzy C-means and C-means. At first, the VA data weresegmented using the Running Damage Extraction (RDE) technique. Then, each segment produced wasanalysed using the strain life approach and global statistical signal values. Finally, the accuracy of each clustering technique wasmeasured based on the OV coefficient index. From the study, the index shows that the Fuzzy C-means technique produced much better clusters rather than the C-mean clustering technique.