Research On Preprocessing And Postprocessing Of Structure Analysis Based On GPU Heterogeneous Platform

With development of computer hardware, attention has been paid on the capacity of parallel computing and floating point computing in computer graphics processor(GPU). Finite element analysis(FEA) based on GPU heterogeneous platform has been applied systematically, however, it is still lacking in preprocessing and post-processing sections for the FEA based on GPU heterogeneous platform. With regard to preprocessing section, this paper contributes to rapid modeling by implementing data extraction procedures.Regarding the post-processing section, visualization system of the FEA based on GPU heterogeneous platform is established and visualization processing is implemented in current paper. This paper includes the following aspects:Research on the preprocessing section of the FEA based on GPU heterogeneous platform: To realize rapid modeling, the Python language is programmed to extract information from the INP file of ABAQUS model, which consists of the node information, element information, material information,section information, the reinforcement information, quality information, load information and boundary information, and then the extracted data is transmitted to the GPU memory so that GPU can call the calculation module and analysis the model. In this paper, an example is presented to verify the accuracy of the data extraction program, in which the extracted data is successfully used in the calculation of the FEA program.Establishment of the visualization system of the FEA based on GPU heterogeneous platform: To meet the requirements for the display of FEA data,the FEA visualization system with a well man-machine interactive function is established on the basis of the object-oriented feature of the Python language integrated with GUI toolbox wx Python. In order to realize the view of the model,the operating functions including translation, rotation and scaling the model are implemented in the visual system. The system utilizes Open GL graphics programming interfaces and Python 2D graphics library for analog finite element data to establish the visualization system.Studies on visualization method of FEA based on GPU heterogeneous platform: For the visualization of FEA results, the Open GL is used to develop the data processing program, research and analysis the algorithm and principle of the contour lines and cloud images, and realize grid model drawing, model deformation drawing, contour lines drawing and cloud drawing. Finally, in orderto draw the index parameters of the FEA results, Python’s 2D graphics library Matplotlib is employed to draw the index parameters by calling the numpy array operation. Comparing with the commercial software, well results are achieved by the visualization method based on heterogeneous platform.