圆筒试验JWL状态方程参数的贝叶斯标定

在研究数值模拟的输入参数引入的不确定性时，通常需要人为给定每个输入参数的概率分布，且输入参数概率分布的选择可能会对分析结果产生直接影响。利用贝叶斯方法标定了圆筒试验JWL状态方程参数，得到了标定参数的估计值和后验分布，并研究了不同统计模型假设对标定参数的估计值和后验分布的影响。贝叶斯后验分布融合了基准试验的试验数据的信息，因此将其作为不确定度量化分析时输入参数的初始概率分布，可以尽量减少分布选择引入的认知不确定性。

Bayesian calibration for parameters of JWL equation of state in cylinder test

Since probability distributions of input parameters are usually assigned subjectively for uncertainty quantification (UQ) in numerical simulations,the selection of distributions may have significant effect on analysis results of UQ.In this paper,to calibrate more precisely the parameters of JWL equation of state in the cylinder test,we proposed to adopt Bayesian methods that provide estimators and posterior distributions of calibration parameters.Further,we investigated the effects of model assumptions on estimators and posterior distributions of calibration parameters.Our study shows that,owing to the information of cylinder tests they contain,the posterior distributions can be used as initial probability distributions of the input parameters in UQ to reduce the epistemic uncertainty.