Stiffness determination of angular-contact ball bearings by using neural network

All angular-contact ball bearings have similar features regarding geometry, mechanism, and structure. The stiffness of this type of bearings can be related to geometry, dimension, and operating conditions by a very complex function. This function involves high order and coupled variables. This study presents this stiffness function for all angular-contact ball bearings by a back-propagation neural network method (BPNN), which is trained by using several (not all) samples. The utility of the BPNN is demonstrated for actual cases. Each are catalogued SKF series angular-contact ball bearings.