FSI Data Interpolation Methods Of Curved Coupling Interface By Dimensionality Reduction

In engineering applications,a variety of fluid-structure interaction(FSI)phenomena are often encountered.Now the weak coupling calculation methods of computational fluid dynamics and computational structural dynamics has become the main approach to solve FSI problem.In order to characterize the coupling between two disciplines,the method of FSI data transfer is the key to realize the FSI analysis.Traditional FSI data interpolation methods ignore the space nonlinearity of coupling surface.Although the FSI data interpolation method based on ISOMAP avoids the influence of space nonlinearity of coupling surface on FSI data transfer,ISOMAP uses the Geodesic distance which increases the complexity of calculation.In this paper,we study the ability and efficiency of dimensionality reduction methods to solve the space nonlinearity problem of coupling surface,and the accuracy of the FSI data interpolation method based on dimensionality reduction.The ability of the 14 kinds of dimensionality reduction methods to unfold the coupling surface with space nonlinearity are tested by the unfolding of cylindrical surface and blade FSI interface,and the 5 kinds of dimensionality reduction methods,ISOMAP,LISOMAP,LLE,HLLE and LTSA,are outstanding.From the times used by 5 kinds of dimensionality reduction methods to unfold blade FSI interface,it can be seen that LLE,HLLE and LTSA are more efficient than ISOMAP and LISOMAP.The fluid pressure of blade FSI interface and nozzle FSI interface are transferred by the FSI data interpolation method based on 5 kinds of dimensionality reduction methods and 4 kinds of traditional FSI data interpolation methods.From the error analysis,we can see that the methods based on dimensionality reduction has higher precision.Considering the ability,efficiency and accuracy of 5 kinds of the FSI data interpolation methods,it is found that the methods based on LLE,HLLE and LTSA are more excellent.