Statistical distribution modelling and parameter identification of the dynamic stress spectrum of a tractor front driven axle

The reliability data from the front driven axle of a four-wheel drive tractor exhibits a multi-peak distribution because of the complex field environment. The aim of this study was to establish an accurate statistical model for the dynamic stress spectrum with mixed distributions. Using rain-flow counting, the stress-time history of the tractor was transformed into the rain-flow matrix with two parameters. The mean and range of the stress data in the rain-flow domain was modelled with the mixed Gaussian and Weibull distributions, respectively. The identification of unknown parameters is an important and here a novel parameter identification method that adopts a genetic algorithm (GA) for rapid searching was proposed with the generalised reduced gradient (GRG) method was applied for accurate identification. The results of finite element analysis (FEA) provide reference for the selection of the measured points. To obtain the dynamic strain signal, a real-time test system was built and a field experiment was carried out. The calculation and evaluation results confirmed that this approach precisely estimated parameters and established a highly accurate mixed model. For the mixed Gaussian and Weibull distributions, the determination coefficients R2 exceed 0.98 and the maximum KS test values were not more than 0.2. Furthermore, this study provided accurate model parameters to determine the stress spectrum for the full-life cycle for use in future research.