Method for forecasting elastic deformation of flexible super-long blade based on GPU (Graphics Processing Unit) cluster

The invention discloses a flexible super-long blade elastic deformation forecasting method based on a GPU cluster in the technical field of forecasting. The method comprises the following steps: constructing a super-large wind turbine generator thin-wall blade surface model by using a self-adaptive radial basis function; carrying out background grid identification and division on the thin-wall blade model by adopting a virtual grid method; reconstructing an air domain variable value of the background virtual grid under each time step propulsion, and applying an object plane boundary condition by adopting an interpolation function; for the large deformation part of the super-long blade, modeling near wall surface flow by adopting a wall surface analytic function, reconstructing a virtual grid variable through a wall surface model so as to apply a sliding speed boundary condition and a shear stress boundary condition, and calculating the elastic deformation of the blade; grid unit information is optimally distr