Font Size: a A A

Study On Precise Regulation And Measurement Method Of Wide-Energy-Range Neutron Spectrum Of Reactor

Posted on:2023-09-25Degree:DoctorType:Dissertation
Country:ChinaCandidate:C L CaoFull Text:PDF
GTID:1522306902954109Subject:Nuclear science and engineering
Abstract/Summary:PDF Full Text Request
Neutron technology is extensively applied in significant fields such as reactor system design,neutronics experiment,and industrial non-destructive testing.The precise regulation and measurement technology of neutron spectrum,that is,obtaining a neutron spectrum that meets specific requirements and realizing the measurement,is the crucial matter of the whole neutron regulation technology.The precise regulation of reactor neutron spectrum affects the economy and safety of reactor design and the reliability of neutronics experiment,etc.Meanwhile,precise measurement is one of the key factors to ensure the accuracy of regulation.The traditional neutron spectrum regulation technology relies on manual experience to iteratively design the regulation module,which is difficult to guarantee the precise regulation.In addition,the research on the online measurement technology of wide-energy-range neutron spectrum also has the technical challenge that is tough to improve the accuracy of the spectrum unfolding,and it is difficult to accurately evaluate the regulated neutron spectrum.In this paper,combining the idea of reverse regulation and deep learning,the research on the precise regulation and measurement method of wide-energy-range neutron spectrum of reactor was carried out.The main contents and innovations include:(1)A reverse regulation method of neutron spectrum on the basis of intelligent correction was proposed,and the difference between the source neutron spectrum and the target neutron spectrum was continuously eliminated through self-adaptive optimization,which improved the accuracy and efficiency of neutron spectrum regulation.The method pre-designed different types of regulation modules and calibrated them with multi-dimensional response matrix through neutron transport simulation,optimizing the arrangement of regulation modules with the differential evolution algorithm,and then used the artificial neural network algorithm to correct the thickness of regulation modules,so as to generate the optimal regulation scheme.Through the verification on the target neutron spectrum,the results showed that the method designed in this paper has higher regulation accuracy and efficiency than the traditional regulation method.(2)Combining neutron spectrum features of reactor and database of neutron spectra,a wide-energy-range neutron spectrum unfolding algorithm of reactor based on deep learning was proposed.For regions with obvious spectral characteristics in reactor,a spectrum unfolding algorithm based on spectral features recognition was constructed,and it established high-quality default spectrum by mining the inherent law between the detector count ratio and the characteristic peak positions that conform to the spectral features of the reactor,thereby improving the accuracy of traditional measurement methods.For regions with unobvious spectral characteristics in reactor,a spectrum unfolding algorithm based on transfer learning was constructed,and it extended the model structure to train deep neural networks,breaking through the limitation of insufficient training data,and realized the adaptive deviation-resistant ability of the unfolded spectrum.The verification outcomes indicated that the proposed method has higher accuracy than the iteration method with traditional default spectrum.In order to verify the application effect of the developed method in this paper in different reactors,the research utilized the neutron spectrum of test blanket module(TBM)of fusion reactor ITER and the core neutron spectrum of fission reactor(RBEC-M)respectively to test the proposed method comprehensively.The results showed that on the ITER TBM spectrum,the regulation method in this paper reduced the MSE by 7.69%compared to the Mock-up experimental spectrum,and the measurement method reduced the mean squared error(MSE)by 13.40%compared to the traditional method.On the RBEC-M core spectrum,the regulation method in this paper reached the magnitude of 1E-06 in MSE,and the measurement method reduced the MSE by 7.1%compared to the traditional method.At the same time,the proposed method in this paper has a lower time complexity,which confirms its application potential in the field of fast and precise regulation and measurement of wide-energy-range neutron spectrum of reactor.
Keywords/Search Tags:Neutron spectrum of reactor, Regulation and measurement, Reverse regulation, Deep learning, Multi-dimensional response matrix
PDF Full Text Request
Related items