Research And Implementation Of Parallel Computing For Cellular Automata Evacuation Simulation

Large-scale densely-populated places have become more and more common in daily life,and the evacuation safety problems caused by them have become one of the typical public safety problems in today's society.Research on evacuation behavior in densely-populated places has important guiding significance for public safety issues.Establishing evacuation simulation model based on pedestrian dynamics is an efficient and low-cost method for evacuation process research.Establishing a high-performance evacuation simulation model has practical value in preventing and dealing with public safety incidents.The current pedestrian dynamics model mainly focuses on the accurate description of evacuation behavior,but neglects the performance of the evacuation model,which makes it difficult to simulate large-scale evacuation scenes.This paper focuses on the performance of the evacuation model.Based on the cellular automated evacuation model,the performance problems of the evacuation model in large-scale scene simulation are solved by parallel computing.In this paper,a cellular automaton evacuation model based on triangular meshing is established.The model subdivides the traditional quadrilateral grid into triangles,so it can provide 14 different evacuation directions for pedestrian.This makes the model superior in expressing the diversity of moving directions of pedestrian,and can also simulate the interaction of evacuation process as vividly as possible in high-density scenes.In order to improve the computational efficiency of the evacuation model,this paper parallelizes the cellular automata evacuation model,so that the evacuation model can improve the computational efficiency by adding processors.This paper considers the influence of different task partitioning methods on performance in parallel algorithm design,and provides different task partitioning methods for different scenarios to achieve the best parallel effect.At the same time,the problem of data conflict between evacuees and inconsistency of iteration schedule in parallel evacuation algorithm is solved efficiently.Experiments show that the results of the parallel evacuation model are consistent with the original algorithm.In terms of efficiency,the parallel evacuation model can easily obtain more than 300% acceleration ratio at a certain computing scale,which is significantly superior to the original algorithm in efficiency.At the same time,the scalability of the parallel model is verified,which means that with the increase of the scale of the problem,the parallel model will get higher performance.Through the simulation of actual cases and comparing with the evacuation simulation software Pathfinder,it can be seen that the evacuation simulation results of the model are in good agreement with expectations,and have an order of magnitude advantage in solving efficiency.At the same time,it reflects the behavior of some pedestrians(such as backing up when blocking).