贝叶斯推断法标定核燃料裂变气体释放模型  

Calibrating the Nuclear Fuel Fission Gas Release Model by Bayesian Inference Method

作　　者：潘昕怿 王业辉 张盼 兰兵 依岩 PAN Xin-yi;WANG Ye-hui;ZHANG Pan;LAN Bing;YI Yan(Nuclear and Radiation Safety Center MEE,Beijing 100082,China)

机构地区：[1]生态环境部核与辐射安全中心,北京100082

出　　处：《核电子学与探测技术》2023年第2期266-270,共5页Nuclear Electronics & Detection Technology

基　　金：国家重点研发计划(2019YFB1900801)资助。

摘　　要：针对模型参数不确定性影响裂变气体释放(FGR)机理模型预测精度的问题,构建了一套贝叶斯标定方法。利用FGR实验测量数据标定晶内气体扩散系数等模型参数并推断其后验概率分布,采用Kriging模型和主成分分析法提高贝叶斯推断效率。分析结果表明,标定后模型FGR计算结果的准确性显著提高,总体均方根误差降低约70%;5个模型参数后验分布标准差相比先验分布均有所减小,进而降低了FGR预测值的不确定度。A Bayesian calibration method is established to address the problem that the uncertainty of the parameters of the fission gas release(FGR)mechanism model affects the accuracy of prediction results.FGR measurement data are used to calibrate model parameters including the intragranular gaseous diffusion coefficient and infer their posterior probability distribution,while using Kriging model and principal component analysis method to improve the efficiency of the Bayesian inference.The analysis results indicate that the calibrated model can significantly improve the FGR calculation accuracy,and the overall root mean square error is reduced by about 70%;the posterior distribution standard deviation of the five model parameters is reduced by varying degree compared with the prior distribution,thereby reducing the uncertainty of the FGR predictions.

关 键 词：裂变气体释放 标定 贝叶斯推断 KRIGING模型 

分 类 号：TL352[核科学技术—核技术及应用]