Peak-valley segmentation algorithm for fatigue time series data

Abstract

This paper presents the peak-valley (PV) segmentation algorithm for the purpose of producing a reliable method of fatigue time series segmentation and statistical segment-by-segment analysis of fatigue damage. The time series were segmented using a piecewise linear representation (PLR) based segmentation algorithm and consecutively the peak-valley (PV) segmentation algorithm. Statistical analysis and fatigue damage calculations were made on each segment and scatter plots were produced based on the relationship between segmental damage and its corresponding kurtosis value. Observations were made on the scatter plots produced by the PV segmentation algorithm to determine the reliability of the data scattering for fatigue data clustering prospects.