Prediction method and sensitivity analysis of the heat generation for an angular contact ball bearing

Latin hypercube experimental design method was used to produce the sample data of structure parameters for the angular contact ball bearing NSK-7015C. The Newton-Raphson method was proposed for solving nonlinear equations composed of the equation of kinematic constraints, the ball equilibrium equations and the balance equations of the inner ring of the bearing. Then, the sample points corresponding the heat generation of the bearings were obtained. A combination method (active learning and Kriging-based Monte-Carlo simulation,AK-MCS) of the Kriging model and Monte Carlo was adopted to establish a function between sample points and the heat generation of the bearing. On this basis, a global sensitivity computational method was used to calculate the influence of structure parameters of the bearing on its heat generation. The results show that the differences between the prediction results of the heat generation for the bearing using AK-MCS method and the quasi-static analysis model of the bearing under the same structure parameters of the bearing are less than 0.003 W. This illustrates that the prediction results of the AK-MCS algorithm are highly accurate. The heat generation of the bearing is the most sensitive to the change of ball diameter,then the inner- and outer-ring raceway diameter take the second place, and the curvature radius of inner- and outer-ring raceway groove are the least. This research may bring forward the theoretical basis for the design optimization of structure parameters and selecting the machining accuracy of the bearing.