We proposed and compared three methods(filter burnup,single energy burnup,and burnup extremum analysis)to build a high-resolution neutronics model for 238Pu production in high-flux reactors.The filter burnup and singl...We proposed and compared three methods(filter burnup,single energy burnup,and burnup extremum analysis)to build a high-resolution neutronics model for 238Pu production in high-flux reactors.The filter burnup and single energy burnup methods have no theoretical approximation and can achieve a spectrum resolution of up to~1 eV,thereby constructing the importance curve and yield curve of the full energy range.The burnup extreme analysis method combines the importance and yield curves to consider the influence of irradiation time on production efficiency,thereby constructing extreme curves.The three curves,which quantify the transmutation rate of the nuclei in each energy region,are of physical significance because they have similar distributions.A high-resolution neutronics model for ^(238)Pu production was established based on these three curves,and its universality and feasibility were proven.The neutronics model can guide the neutron spectrum optimization and improve the yield of ^(238)Pu by up to 18.81%.The neutronics model revealed the law of nuclei transmutation in all energy regions with high spectrum resolution,thus providing theoretical support for high-flux reactor design and irradiation production of ^(238)Pu.展开更多
A passive neutron multiplicity measurement device,FH-NCM/S1,based on field-programmable gate arrays(FPGAs),is developed specifically for measuring the mass of plutonium-240(^(240)Pu)in mixed oxide fuel.FH-NCM/S1 adopt...A passive neutron multiplicity measurement device,FH-NCM/S1,based on field-programmable gate arrays(FPGAs),is developed specifically for measuring the mass of plutonium-240(^(240)Pu)in mixed oxide fuel.FH-NCM/S1 adopts an inte-grated approach,combining the shift register analysis mode with the pulse-position timestamp mode using an FPGA.The optimal effective length of the^(3)He neutron detector was determined to be 30 cm,and the thickness of the graphite reflector was ascertained to be 15 cm through MCNP simulations.After fabricating the device,calibration measurements were per-formed using a^(252)Cf neutron source;a detection efficiency of 43.07%and detector die-away time of 55.79μs were observed.Nine samples of plutonium oxide were measured under identical conditions using the FH-NCM/S1 in shift register analysis mode and a plutonium waste multiplicity counter.The obtained double rates underwent corrections for detection efficiency(ε)and double gate fraction(f_(d)),resulting in corrected double rates(D_(c)),which were used to validate the accuracy of the shift register analysis mode.Furthermore,the device exhibited fluctuations in the measurement results,and within a single 20 s measurement,these fluctuations remained below 10%.After 30 cycles,the relative error in the mass of^(240)Pu was less than 5%.Finally,correlation calculations confirmed the robust consistency of both measurement modes.This study holds specific significance for the subsequent design and development of neutron multiplicity devices.展开更多
识别虚假评论有着重要的理论意义与现实价值.先前工作集中于启发式策略和传统的全监督学习算法.最近研究表明:人类无法通过先验知识有效识别虚假评论,手工标注的数据集必定存在一定数量的误例,因此简单使用传统的全监督学习算法识别虚...识别虚假评论有着重要的理论意义与现实价值.先前工作集中于启发式策略和传统的全监督学习算法.最近研究表明:人类无法通过先验知识有效识别虚假评论,手工标注的数据集必定存在一定数量的误例,因此简单使用传统的全监督学习算法识别虚假评论并不合理.容易被错误标注的样例称为间谍样例,如何确定这些样例的类别标签将直接影响分类器的性能.基于少量的真实评论和大量的未标注评论,提出一种创新的PU(positive and unlabeled)学习框架来识别虚假评论.首先,从无标注数据集中识别出少量可信度较高的负例.其次,通过整合LDA(latent Dirichlet allocation)和K-means,分别计算出多个代表性的正例和负例.接着,基于狄利克雷过程混合模型(Dirichlet process mixture model,DPMM),对所有间谍样例进行聚类,混合种群性和个体性策略来确定间谍样例的类别标签.最后,多核学习算法被用来训练最终的分类器.数值实验证实了所提算法的有效性,超过当前的基准.展开更多
基金supported by Natural Science Foundation of China (No. 12305190)Lingchuang Research Project of China National Nuclear Corporation (CNNC)the Science and Technology on Reactor System Design Technology Laboratory
文摘We proposed and compared three methods(filter burnup,single energy burnup,and burnup extremum analysis)to build a high-resolution neutronics model for 238Pu production in high-flux reactors.The filter burnup and single energy burnup methods have no theoretical approximation and can achieve a spectrum resolution of up to~1 eV,thereby constructing the importance curve and yield curve of the full energy range.The burnup extreme analysis method combines the importance and yield curves to consider the influence of irradiation time on production efficiency,thereby constructing extreme curves.The three curves,which quantify the transmutation rate of the nuclei in each energy region,are of physical significance because they have similar distributions.A high-resolution neutronics model for ^(238)Pu production was established based on these three curves,and its universality and feasibility were proven.The neutronics model can guide the neutron spectrum optimization and improve the yield of ^(238)Pu by up to 18.81%.The neutronics model revealed the law of nuclei transmutation in all energy regions with high spectrum resolution,thus providing theoretical support for high-flux reactor design and irradiation production of ^(238)Pu.
基金supported by the National Natural Science Foundation of China(No.42374226)Natural Science Foundation of Jiangxi Province(Nos.20232BAB201043 and 20232BCJ23006)+1 种基金a sub-project of the nuclear energy development project of the China National Defense Science and Industry Bureau‘n-γfusion logging method theory research’(No.20201192-01)the Fundamental Science on Radioactive Geology and Exploration Technology Laboratory(No.2022RGET20)。
文摘A passive neutron multiplicity measurement device,FH-NCM/S1,based on field-programmable gate arrays(FPGAs),is developed specifically for measuring the mass of plutonium-240(^(240)Pu)in mixed oxide fuel.FH-NCM/S1 adopts an inte-grated approach,combining the shift register analysis mode with the pulse-position timestamp mode using an FPGA.The optimal effective length of the^(3)He neutron detector was determined to be 30 cm,and the thickness of the graphite reflector was ascertained to be 15 cm through MCNP simulations.After fabricating the device,calibration measurements were per-formed using a^(252)Cf neutron source;a detection efficiency of 43.07%and detector die-away time of 55.79μs were observed.Nine samples of plutonium oxide were measured under identical conditions using the FH-NCM/S1 in shift register analysis mode and a plutonium waste multiplicity counter.The obtained double rates underwent corrections for detection efficiency(ε)and double gate fraction(f_(d)),resulting in corrected double rates(D_(c)),which were used to validate the accuracy of the shift register analysis mode.Furthermore,the device exhibited fluctuations in the measurement results,and within a single 20 s measurement,these fluctuations remained below 10%.After 30 cycles,the relative error in the mass of^(240)Pu was less than 5%.Finally,correlation calculations confirmed the robust consistency of both measurement modes.This study holds specific significance for the subsequent design and development of neutron multiplicity devices.
文摘识别虚假评论有着重要的理论意义与现实价值.先前工作集中于启发式策略和传统的全监督学习算法.最近研究表明:人类无法通过先验知识有效识别虚假评论,手工标注的数据集必定存在一定数量的误例,因此简单使用传统的全监督学习算法识别虚假评论并不合理.容易被错误标注的样例称为间谍样例,如何确定这些样例的类别标签将直接影响分类器的性能.基于少量的真实评论和大量的未标注评论,提出一种创新的PU(positive and unlabeled)学习框架来识别虚假评论.首先,从无标注数据集中识别出少量可信度较高的负例.其次,通过整合LDA(latent Dirichlet allocation)和K-means,分别计算出多个代表性的正例和负例.接着,基于狄利克雷过程混合模型(Dirichlet process mixture model,DPMM),对所有间谍样例进行聚类,混合种群性和个体性策略来确定间谍样例的类别标签.最后,多核学习算法被用来训练最终的分类器.数值实验证实了所提算法的有效性,超过当前的基准.
