基于回归分析模型与随机过程模型的硅橡胶泡沫应力松弛预测对比

Comparison of Stress Relaxation Prediction of Silicone Rubber Foam Based on Regression Analysis Model and Stochastic Process Models

本文针对硅橡胶泡沫的应力松弛实验数据,基于二阶Prony级数,采用非线性回归分析方法(曲线拟合法)、基于Wiener过程的贝叶斯方法、基于Gamma过程的贝叶斯方法进行参数估计从而分别建立预测模型,通过对比三种建模方法的预测值与实验值的偏差,比较了三种建模方法的效果。结果发现,采用曲线拟合的方法进行参数估计,其预测偏差要大于基于Wiener、Gamma随机过程的贝叶斯方法,表明由于考虑了性能退化过程存在的不可避免的随机性,随机过程模型可以更好地预测硅橡胶泡沫的应力松弛趋势,同时由于应力松弛过程是单调增加的,基于非负的Gamma过程的随机过程模型的预测偏差要小于基于Wiener过程的模型。本文工作可为基于高分子材料性能退化数据的退化预测模型的建模提供一种较好的参数估计方法,有助于提高预测结果的准确性。

In this study,we tried to deal with the experimental data of stress relaxation of silicone rubber foam.Based on the second-order Prony series,the nonliner regression analysis method(curve fitting method),the Bayesian method based on the Wiener process and the Bayesian method based on the Gamma process were used to estimate the parameters,and the prediction models are established respectively.By comparing the deviation between the predicted values and experimental values of the three modeling methods,the effects of the three modeling methods are compared.The results showed that the prediction deviation of the model based on the regression analysis was larger than the ones established by the Bayesian method based on the Wiener and Gamma processes,which may indicate that the stochastic process models could better predict the stress relaxation tendency of silicone rubber.The non-negative Gamma process,suitable for the nature of monotone increasing of a stress relaxation process,may contribute to its less deviation than the Wiener process.Our work provided a better way for the parameters estimation of degradation model based on the performance degradation data of polymer materials,contributing to better prediction.