Distributions and fitting of stress amplitude spectra for railway vehicle structures
Stress spectra are essential for estimating the structural fatigue and the fatigue life of railway vehicle structures. However, the data grouping and distribution fitting approaches employed by previous empirical data processing methods are excessively coarse and contain distinctive defects. To address these shortcomings, general theoretical distributions are herein corrected according to the requirements presented by lower stress amplitude data bounds, and converted into a Weibull distribution, truncated lognormal distribution, and truncated normal distribution for comparison. In addition, an innovative distribution consisting of a combined Weibull distribution and truncated lognormal distribution is proposed. The results of fitting to measured stress spectra suggest that the combined distribution has remarkable advantages. Stress spectrum extrapolation and fatigue damage calculation verify that the stress spectrum derived from the combined distribution provides the closest match of all distributions considered to that obtained from the measured spectrum, and therefore best represents the stress distribution at critical points during the practical operation of railway vehicles.