Research And Simulated Implementation Of Smoke Modeling Method Based On Big Scene Environment

Nowadays,smoke simulation is an important part of current natural environment simulation.Eliminating artificial traces and enhancing the authenticity of smoke simulation have become the main problems faced by researchers.In this thesis,the momentum based N-S fluid movement model is changed,and a double precision smoke particle system is constructed to improve the real-time performance of smoke simulation by using the vorticity based smoke diffusion model.The low precision particle system mainly studies the basic diffusion law of smoke particles.In view of the specific influence of the external force on the smoke particles,the basic motion trajectory of the smoke particles is depicted,and provides the corresponding physical information for high-precision particle system.After constructing a low precision smoke particle system,a high precision smoke particle system is constructed by grid diffusion.At the same time,the convection term and the diffusion term are solved in the constructed uniform grid.The constructed homogeneous grid is used to store the eddy information,and the improved semi-Lagrangian method is used to solve the grid interpolation.The influence of the boundary conditions on the smoke vorticity particles is studied,and the authenticity of the smoke simulation is increased.Taking into account the similarity of the trajectories of flame particles and smoke particles,the same diffusion model is chosen to simulate the flame particles.The threshold temperature of flame to smoke is calculated by the equation of oxidation-reduction reaction equation.Considering the relationship between temperature and density,the physical information of temperature is combined with the N-S diffusion equation,which further increases the authenticity of smoke simulation.Then the BP neural network is designed to fuse the features of the extracted steam.We detailed analyze the design principles and selection of parameters.The experiment results show that the steam detection algorithm in this thesis can detect the position of the steam in the video accurately,real-timely and effectively in the process of steamer filling.Due to the large amount of computation,the real-time nature of smoke simulation is severely affected.In order to increase the visual experience,multi-threaded processing is used in this paper to find independent and time-consuming modules.By using multi-core computer processing techniques,under the conditions of ensuring the authenticity of smoke to a great extent to improve the real-time smoke simulation.