Probabilistic assessment of heavy-haul railway track using multi-gene genetic programming

Abstract

This study presented a probabilistic assessment of heavy-haul railway track using a high-performance computational model called multi-gene genetic programming (MGGP). A reliability analysis (RA) method based on MGGP and the first-order second-moment method (FOSM) has been proposed in this study. First, GP was used to map the implicit performance functions; therefore, arriving at GP-based explicit performance functions. Subsequently, the developed GP model was used to perform RA of a soil slope of heavy-haul railway track under both seismic and non-seismic conditions. Using the FOSM, soil uncertainties were mapped based on the concepts of probability theory and statistics, and a ready-made expression was developed. Simulated results demonstrate that the GP-based FOSM approach can predict the probability of failure (POF) of slope with rational accuracy. The probabilistic analysis against bearing capacity failure was also investigated in this study to ensure serviceability of the soil slope. Based on the outcomes, it can be deduced that the coefficient of variation of soil properties affects the POF of slope significantly. With the aid of the developed expression, the POF of the soil slope of heavy-haul railway track can be assessed rationally and efficiently.