Machine learning calculation model for hydrodynamic lubrication characteristics of a miter gate bottom pivot.

The bottom pivot is a vital support device in the miter gate but often subject to poor lubrication and wear failures. Calculating the hydrodynamic lubrication characteristics of the bottom pivot is a complex three-dimensional (3D) problem, and most of existing models adopt simplified assumptions to reduce the calculation difficulty. To solve this issue, this work develops a 3D model to calculate the hydrodynamic lubrication characteristics of the miter gate bottom pivot. The finite difference method is used to solve the oil film thickness and pressure distribution based on the spherical coordinates Reynolds equation. The component forces in three directions are calculated from the pressure distribution and compared with the theoretical values to generate the calculation difference. Then, the genetic algorithm (GA) is used to minimize the difference to determine the optimal initial parameters for the 3D model. The analysis results show that the calculation accuracy can be significantly improved by using the optimal initial model parameters. When our initial pressure is 5.64MPa, the results meet the engineering accuracy requirements. [ABSTRACT FROM AUTHOR]