Research On Key Technologies And Application Of Motion Simulation For Large Scale Virtual Factory Scene

With the proposal of "making 2025 in China" plan,the production mode of the traditional factory is changing to the direction of intelligent manufacturing.Unmanned factory plant is the basic manifestation of intelligent manufacturing,and the simulation of large-scale scene in virtual factory is an important technical support for realize unmanned production.The 3D virtual factory simulation system is an important bridge for people to communicate with production information,and the three-dimensional space can more intuitively demonstrate the plant's real-time production status.With the everincreasing scale of plants and the increasing number of equipment,the amount of data that virtual simulation systems needs to deal with is also expanding.In order to solve the problem of large amount of computation and low performance of simulation system caused by model data in large scale scenes,in this paper the acceleration technology based on GPU and culling algorithms based on object visibility are studied,the main contents are as follows:(1)The motion simulation of virtual device based on GPU is completed.By measuring the actual equipment size and position of the plant,the drawing of threedimensional model of the plant and the internal equipment is completed.By analyzing the three-dimensional model of the plant equipment,the kinematics equations in the OpenGL world coordinate system are solved,and the programmable pipeline data interface is designed.According to the characteristics of the plant equipment,multiple drawing methods of the same model device driven by different real-time data are proposed.By programming GPU programmable Pipeline shader,the OpenGL instantiation technology is used to realize the drawing of batches of static objects in the plant.(2)The culling algorithms based on bounding box and octree structure is designed.The algorithm of reading,storing and redrawing 3D models in OBJ format is designed,and the bounding boxes of the equipment are solved by using parsed model data.According to the location characteristics of plant equipment and the data of bounding box,the spatial segmentation based on octree structure is carried out.The updating process of hierarchical bounding box and octree structure is realized by using external interactive data of simulation system.Finally,based on the relationship between the node of Octree structure and the spatial position of the scene,the design of the invisible object culling algorithm based on bounding box and octree structure is completed.(3)Design and performance analysis of virtual simulation system.Taking the fixed pipeline as a benchmark,the performance of the completed simulation system is tested,and the efficiency of the drawing speed of the virtual scene is verified by the GPU programmable pipeline and the culling algorithm.From the number of test frames,it can be seen that the main influencing factors affecting the operating efficiency of the simulation system are the large amount of texture data in the virtual scene.Finally,the data acquisition software is integrated to verify the reliability of the simulation system on the test machine.