Prediction of the reaction forces of spiral-groove gas journal bearings by artificial neural network regression models

This paper presents neural network regression models for predicting the nonlinear static and linearized dynamic reaction forces of spiral grooved gas journal bearings. The partial differential equations (PDEs) are sampled, based on a full factorial randomly spaced parameter set. Feed-forward (FNN) architectures developed modeling PDEs therefore replacing time-consuming discrete iterative solution procedure used to this date. A significant speed-up factor >103 in computation time is achieved, compared solving PDE numerically. Furthermore, FNN allows multi-dimensional interpolation, which makes global system optimization easily possible. demonstrated by real-case rotordynamic optimization. By using meta-models, complete reduction 300 achieved.