A new non-Gaussian to Gaussian transformation approach for fatigue life analysis applied to long-term wind speed data

Purpose The main objective of this study is to develop an efficient non-Gaussian to Gaussian transformation method for extremely non-Gaussian data. The second objective is to compare the proposed new method and eight common transformation methods in how close the kurtosis and skewness of their resulting transformed data matched Gaussian values. The third objective was evaluating the fatigue lifetime of a structure by using these transformed data and seeing how closely they match the results of the time domain (TD) fatigue analysis using rainflow cycle counting. Design/methodology/approach The new method is based on using the Slifker and Nelder–Mead algorithms consecutively in the Johnson B transformation. The initial estimate of the Johnson parameters obtained by the Slifker algorithm was iteratively improved by the Nelder–Mead's Downhill Simplex optimization algorithm to generate new parameters. National Oceanic and Atmospheric Administration of the United States of America wind data were used to calculate the stress on a traffic sign. The stress was transformed to Gaussian form by the nine methods. The fatigue lifetime of the structure was calculated using the transformed data sets and correction factors from literature and then compared with the prediction of the TD method using rainflow cycle counting. Findings Nelder–Mead optimization resulted in a distribution with exact Gaussian values of skewness and kurtosis. The comparison of the new method with the TD method with rainflow cycle counting showed that the fatigue damage result was the closest match to the TD result. Originality/value This is the first time the Johnson transform has been used with parameters calculated by the Nelder–Mead optimization algorithm for fatigue life estimation. The results of eight transformation methods and the new method were compared for the first time.