Neural network modeling of hydrodynamics processes in the centrifugal pump and oil pipeline

摘要

Artificial neural network using for hydrodynamics processes studying is presented by two fundamentally different approaches. The first one is the neural network using for the direct differential hydrodynamics equations solution. Theses equations describe the 2D and 3D turbulent isothermal flow of the viscous incompressible liquid in centrifugal pump flowing area model. Neural network solution results of hydrodynamic equations for the computational zone that consists of two sub-domains are given below. One of these is rotating, and the second one is immobile. In this case at the neural network algorithm realization it is not required to specify the conjugate conditions at the two sub-domains border. The second approach consist in neural network structures application for the computational experiment results approximation obtained after using of traditional methods of computational hydrodynamics and for obtaining of hydrodynamic processes multifactor approximation models. The present approach is illustrated by the hydrodynamics processes neural network modeling in pipeline in the case of medium leakage through the wall hole