Research On Simplif Ication And Dispatching Technologies Of Large CAE Data For Realtime Visualization Analysis

CAE technology has been increasingly popular in the modern manufacturing industry, as an important component of the industrial product design, it has played an important role in the quality, life, performance and cost in product design. For the many forms and huge CAE data, traditional numerical processing and formulas chart methods, it is far from sufficient to undertake analysis of data results, visual analysis can help designers to extract meaningful features and results from the complex data quickly and efficiently through intuitive visual images. Visualization post-processing has become the primary means of awareness and understanding of CAE data intuitively and vividly.With the complex of the object for analysis and the improvement of calculation accuracy (such as CAE calculation on supper computer), the scale of data calculated is increasing, GB-level CAE data is common. Limited to computer performance, the method that read all data into memory and then draw can no longer meet the needs of visualization post-processing. Visualization post-processing for large-sale CAE Data has become a bottleneck in the development of CAE technology.Based on the research of visualization methods for large-sale data and the deficiencies, simplified method for large-sale CAE multi-resolution data model based on out-of-core and viewpoint-based real-time data scheduling method are proposed, it provides an effective way of data processing for visual analysis of large-sale CAE data.Firstly, after analyzing the features of large-sale CAE data, multi-resolution model of large-sale CAE data based on out-of-core generation method is proposed, which can divide the data adaptively using non-uniform binary tree and solve the problem that can not read all the data to the cache. The method builds a multi-level data indexing tree in memory. Then the sub-block data are simplified with the grid vertex. The simplification is bottom-up from the bottom level of data indexing tree and generates multi-resolution data, providing effective data organization for data scheduling of visualization.Secondly, based on the current status that CAE data have many different types, unified data conversion program is designed for CAE data of multiple types. After the CAE data input, the program identifies the data type automatically and calls the corresponding interface plug-in to read in the CAE data.Thirdly, viewpoint-based scheduling method for large-sale CAE data visualization is proposed, which combined fast Algorithm of eliminating visible region can scheduling out-of-core data of the appropriate level and accuracy to the memory for visualization. The levels of data accuracy are adjusted automatically according to the frame rate, so we can find a balance between quality and speed. Multithread of reading out-of-core data is using to speed up the data schedule.Thirdly, based on the current status that CAE data have many different types, unified data conversion program is designed for CAE data of multiple types. After the CAE data input, the program identifies the data type automatically and calls the corresponding interface plug-in to read in the CAE data. .Finally, on the basis of the above method, large-scale CAE data simplification and Scheduling System is developed. The correctness and effectiveness of the method described in this paper is verified using typical CAE data with the visual platform VRFlier.