The fatigue and fracture of gears in Power Split Device (PSD) can be sensitive to stochastic load which is often difficult to be measured and predicted. Due to the load randomness associated with the mode-switch in a hybrid car, non-parametric extrapolation (NPE) method can be applied in the load spectrum compiling of PSD. In this paper, a kernel selection method based on the multiple-criteria decision making method (MCDM) was presented to improve the accuracy of the kernel density estimation. According to the dynamic and contact analysis of gears, a short-term bending stress spectrum was calculated to provide the reference for the fatigue design. Based on the NPE method considering the kernel selection, a whole-life load spectrum was compiled to supply the input data for the fatigue life test of PSD.

This content is only available via PDF.
You do not currently have access to this content.