LAUSR

Abstract A novel fourth-order linear moment (L-moment) reliability method considering linear correlated (L-correlated) random variables is developed to improve accuracy and applicability on the premise of ensuring FORM's computational efficiency. For accomplishing this task, a novel normal transformation's implementation based on L-moments and L-correlations is established, and the complete monotonic expressions of Pearson correlation coefficient is obtained. Additionally, to guarantee that the specified L-correlation coefficient is valid, the suitable regions of L-correlation coefficient are identified based on the discriminant analysis. Then, the normal transformations are integrated in the FORM to evaluate the reliability of mechanical system with L-correlated variables. Several numerical examples are presented to demonstrate the accuracy, efficiency and applicability of the proposed method. Compared with FORM, the proposed method is improved by taking advantage of the fourth-order L-moments.