Font Size: a A A

Concentration Field Reconstruction Based On Dimension Reduction And CFD Simulation

Posted on:2023-06-26Degree:MasterType:Thesis
Country:ChinaCandidate:C Q WangFull Text:PDF
GTID:2542306941969689Subject:Power Engineering and Engineering Thermophysics
Abstract/Summary:PDF Full Text Request
As an important measurement parameter,concentration data are widely used in the operation and control of power plants.However,complex mechanical structures exist in some electrical equipment,which makes it difficult for sensors to directly measure the concentration of components.In order to solve this problem,this paper proposes a concentration reconstruction method based on CFD simulation and dimension reduction algorithm.This method can realize the full-field measurement of concentration value by measuring the concentration value in the area where the sensor is easy to be arranged.Firstly,the CFD software ANASYS Fluent was used to simulate the leakage process of the tube bundle,and the concentration distribution of the flow field under various working conditions was calculated to provide a sample database for the reconstruction algorithm.Then the paper adopts different dimension reduction algorithms for different dimensions.For the reconstruction of two-dimensional concentration distribution,principal component analysis(PCA)and non-negative matrix factorization(NMF)were used to extract the basis matrix from the sample database,and then the concentration distribution of the whole flow field was obtained by a small amount of concentration measurement.For three-dimensional concentration distribution reconstruction,PC A and Tucker decomposition are used to extract the basis matrix and core tensor from the sample database,and then a small number of measurement point data are used to complete the concentration field reconstruction.Then,the reconstruction error is used to evaluate whether these algorithms can be used in 2D and 3D concentration reconstruction.After determining the appropriate reconstruction algorithm,the influence of the parameters in the algorithm,such as the number of base vectors,the number of measurement points and the measuring point location on the reconstruction error is analyzed,and the accuracy of the reconstruction algorithm is optimized through the analysis results.Finally,the leakage experiment is performed to verify the practicability of the concentration reconstruction algorithm.The results of numerical simulation and leakage experiment show that both PCA and Tucker decomposition can reconstruct the concentration distribution well.Then image processing technology is combined with the concentration reconstruction algorithm to reconstruct the concentration distribution of the whole flow field by taking images of some areas of the flow field,which provides a new concentration measurement method for the flow field with complex mechanical structures.
Keywords/Search Tags:concentration distribution reconstruction, computational fluid dynamics, principal component analysis, non-negative matrix factorization, Tucker dec-omposition, image processing technology
PDF Full Text Request
Related items