Stochastic Performance of Journal Bearing With Two-Layered Porous Bush—A Machine Learning Approach

This investigation presents the deterministic and stochastic responses of the journal bearing with a two-layered porous bush. Pressure equations in the porous layers and modified Reynolds equations in the clearance region are governed by the finite difference method (FDM). Stochastic analysis based on Monte Carlo simulation (MCS) is used to investigate the effect of random variation in input parameters caused by uncertain operating conditions, improper installations, and manufacturing imperfections. In order to enhance computational efficiency, this probabilistic study is conducted in conjunction with the machine learning (ML) model based on the support vector machine (SVM) algorithm. The uncertainty in the bearing responses is presented in the form of the probability density function (PDF), considering both the independent and combined effect of the stochastically varied input parameters. Graphical illustration of the data-driven sensitivity represents the relative significance of each input parameter affecting the steady-state responses of the journal bearing with two-layered porous bush. The findings of the present study reveal that the stochastic variations in the input parameters have a profound influence on the operational characteristics of the porous bearing. The outcome of the present study will be helpful in deciding the operational regime of the porous bearing under the practically relevant stochastic environment.