Study on creep rate prediction model of compacted graphite cast iron under wide range of temperature and stress based on PHNN
MA Tian
JING Guoxi
XU Hongjing
TAO Shuai
DONG Changlong
HUANG Lirong
MA Teng
Abstract：[Objective]Aiming at the problem that existing models for predicting the minimum creep rate of compacted graphite cast iron under wide temperature and stress ranges still have large errors,optimization research on the prediction method was carried out to improve prediction accuracy and expand the application scenarios of the physical hierarchical neural network.[Methods]Firstly,based on uniaxial tensile creep test data of compacted graphite cast iron cylinder head material under 450-550℃and 100-300 MPa,the influence law of temperature and stress on the minimum creep rate was analyzed,and the core influencing factors were clarified;secondly,a physical hierarchical neural network prediction model adapted to creep test characteristics was established,with a hierarchical structure of composite layer and stress layer constructed;thirdly,the summation form creep constitutive model was adopted as the control,and model parameter identification was completed by simulated annealing algorithm;finally,the prediction effect comparison and quantitative accuracy evaluation of the two models were completed.[Results]The results show that the creep properties of compacted graphite cast iron under wide working conditions show significant dispersion,and the influence of temperature on its creep damage is higher than that of stress.The established model can constrain all predicted values of the minimum creep rate within the 2-fold error band of the test values,and the prediction accuracy is greatly improved compared with the 3-fold error band of the control model.This model can effectively adapt to creep rate prediction under wide working conditions,expand its application scope,and provide reference for the analysis of high-temperature creep properties of compacted graphite cast iron.
Keywords：Compacted graphite cast ironWide range of temperaturePhysical hierarchical neural networkCreep constitutive modelMinimum creep rate
Publication Date：2026-04-30
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：7( 15-21 )