Service reliability of a heavy-haul wagon coupler body based on the SMOTE-Bootstrap-Bayes method

Abstract To obtain structural service reliability for a small sample of failure data, a reliability analysis method for a small sample of failure data based on the SMOTE (Synthetic Minority Oversampling Technique) algorithm, Bootstrap method, Bayesian estimation and hypothetical distribution check was proposed, and the feasibility of the method was verified. Taking the coupler body of a heavy-haul wagon as the research object, first, the fatigue bench test was carried out, and a small sample of the fatigue failure data for the coupler body under the given load spectrum was obtained. Then, the reliability analysis method for a small sample of failure data was used to analyze the coupler body failure data, and the curve of the coupler body reliability with failure mileage was finally obtained. The results show that the accuracy of the reliability curve based on the small sample analysis method is significantly higher than that based on sample size expansion and the empirical reliability of a small sample derived from random sampling based on an average rank calculation. When the reliability is 50%, the service mileage of the coupler body is approximately 2.08 million km; when the reliability is 95%, the service mileage is approximately 1.38 million km; and when the reliability is 99%, the service mileage is approximately 1.15 million km. The service reliability analysis method for a coupler body of a heavy-haul wagon based on the SMOTE-Bootstrap-Bayesian estimation has important theoretical and application value for improving the utilization rate of test data, reducing test costs and effectively evaluating the fatigue life of coupler bodies.