A GPU-based Simulation Approach Of Neutral Gas Dispersion And Emergent Rescue Research

Because of dangerous gases always being flammable, explosive, toxic, corrosive and so on, the neutral gas leak can cause great damage to people's life, property, the national economy and environmental pollution. Therefore, the gas leak dispersion and emergent rescue become a hot spot. When multiple sources of gas leakage spread, as the executions are very large, common computer can not be real-time response of the spread of gas. In order to real-time prediction of gas dispersion range, we propose a parallel multiple sources Gaussian diffusion model based on GPU-acceleration, which maps parallel algorithm to thread block executing on GPU through the compute unified device architecture(CUDA). The analytical results demonstrate that gas dispersion simulation using this algorithm can meet the demands of time under sudden gas dispersion.In emergent rescue aspect after the neutral gas leak, as the result of emergent rescue algorithm is time-consuming, which cause great damage to people's life, property and the national economy. In order to improve the performance of dynamic programming algorithm while solving the large scale single shortest path problem under incomplete information, we propose a parallel DP algorithm based on GPU-acceleration, which maps parallel DP algorithm to thread block executing on GPU through the compute unified device architecture(CUDA). The analytical results demonstrate that, when we solve the single shortest path problem under incomplete information, the method we proposed increases the scale of problem, speeds up its execution and provide a feasible parallel DP solution for ordinary users.