Numerical Simulation Analysis of Non-Contact Mechanical Seal Performance based on Neural Network

Abstract

Aiming at the problem that the traditional numerical simulation analysis method of sealing performance does not fully consider the sealing parameters, the modeling deviation is large, and the error of the numerical simulation analysis is increased. A neural network-based non-contact mechanical seal performance numerical simulation method is studied. Taking the spiral groove of the dry gas seal type as the research object, the geometric model of the dry gas seal is established and the model is meshed. Calculate the performance parameters of the spiral groove dry gas seal, and train the neural network to realize the numerical simulation of the non-contact mechanical seal performance. Constructing a comparative test with the traditional method verifies that the numerical simulation method using neural network effectively improves the accuracy of the numerical simulation and is effective.