Research On Reactor Core Power Modeling And Reconstruction

The reactor core power and its distribution as the most important monitoring parameters of the core measurement are directly related to the power control of the reactor and the safety of the reactor.At present,large-sized commercial PWRs have a core power distribution reconstruction program based on neutron in-core detectors,and the axial power distribution of the core is usually reconstructed using ex-core neutron detection data.However,for small modular reactors,due to its integrated compact layout,too many detectors cannot be arranged inside the core.Therefore,the real-time reconstruction of the three-dimensional power distribution of the core must rely on the ex-core neutron detection data.Compared with large PWRs,small and medium-sized reactors have a more share of peripheral assembly in the core,so it is more feasible for small and medium-sized reactors to reconstruct the three-dimensional power distribution of the core based on ex-core detection data.To this end,this paper takes a small sized PWR as the object,and conducts in-depth research on the core power distribution reconstruction method.Firstly,a refined physical model of the core of a small-sized PWR was built based on MCNP5,and the reliability of the model was verified by MCNP5 critical calculation under four special core conditions,On this basis,the three-dimensional power distribution of the core under 16 different operating conditions and the corresponding counts of neutron detectors ex-core were calculated by critical source simulation.Secondly,the core nodes are numbered,and the response relationship called transfer matrix between the power of the nodes and the ex-core detector counts is constructed.Then,the expression of the transmission matrix is derived based on the least square method.On this basis,12 of the 16 sets of data are used to find a stable solution of the transfer matrix using truncated singular value decomposition regularization method,and the other 4 sets of data are used for numerical verification.The verification results show that the power distribution reconstruction error based on the transfer matrix method is large,and the average root mean square error is more than 10%.Especially for the nodes of internal assembly and the assembly where the action control rod is located,the relative error of the power reconstruction value is as high as more than 30%.Finally,in view of the large error of the transfer matrix method which is a linear reconstruction method,an intelligent nonlinear reconstruction method based on radial basis function neural network is proposed.Use 12 of the 16 sets of data to train the neural network,and the other 4 sets of data for numerical verification.The verification results show that the average root mean square error of the reconstruction value is 2.24%.The relative error of the reconstructed value of all the nodal is below 10%,and most of them is less than 5%.Both the overall reconstruction effect and the partial reconstruction effect of the reconstruction method based on radial basis function neural network are far superior to the transfer matrix method.The methods and results obtained in this research have a certain reference value for reconstructing the power distribution of the core by using ex-core neutron detection data.