Prediction of rotary seal leakage rate under the influence of stress relaxation based on G A-PSO-BP hybrid algorithm

Abstract Rotary combination seals of composite impactors are susceptible to failure and relaxation in high-pressure, high-temperature deep-well environments. The aim of this study is to analyze the effect of stress relaxation on the performance of the seals and to assess their performance under different operating conditions. The study adopts GA-PSO-BP algorithm to predict the leakage, simulates the contact pressure change of the sealing surface before and after stress relaxation by Abqus software, and calculates the leakage rate by using MATLAB software in combination with the thermoelastic flow lubrication model and the finite element analysis data, to establish the prediction model dataset. The study also optimised the neural network prediction model and verified the accuracy of the model experimentally. The results show that increasing pressure leads to an increase in contact pressure, but relaxation increases contact length and decreases maximum pressure, and increasing compression increases and then decreases contact pressure and increases the peak value, but relaxation decreases the peak value and increases the risk of leakage. The increase in speed and temperature both lead to an increase in leakage rate, and the GA-PSO-BP model predicts more accurately the rotary combination seals with a higher leakage rate after considering the stress relaxation.