Fast Node Search Algorithm Research In The Solution Process Of Fluid Dynamics SPH Method

The Smoothed Particle Hydrodynamics(SPH)method has significant advantages in dealing with free surface large deformation motion,complex boundary flow problems,and fluid flow in multi-connected domains.However,as the number of particles increases during the simulation,the efficiency of particle searching gradually decreases.Therefore,improving the efficiency of searching for node particles is crucial in fluid simulation and is of great significance for achieving fluid simulation.To address this issue,this paper studies a fast parallel search algorithm for nodes based on the fluid mechanics SPH method.Firstly,a graph-based SPH node fast search algorithm is proposed,which includes processes such as particle prediction,neighborhood particle number balance,boundary particle detection,particle-node graph construction,and particle neighborhood search.At the beginning of particle operations,the region is divided by particle prediction and neighborhood particle balance to effectively improve the efficiency of particle calculation.Then,the particle-node graph is constructed to determine a reasonable connection strategy and particle graph update strategy to establish an effective data structure for subsequent particle searches.Finally,the particle search is performed by modifying it through the corresponding graph theory algorithm.This method can effectively reduce the time complexity of neighborhood particle search and provide powerful support for fluid mechanics SPH method computation.Secondly,based on the graph-based search algorithm,a parallel search algorithm based on particle region division is proposed.In order to solve the problem of low efficiency of the fluid mechanics SPH method node search algorithm,this paper uses parallel computing to improve the computational efficiency inthe search algorithm process.When dividing the particle region,particles are evenly divided according to the number of particles and assigned to different threads for calculation.After the region division is completed,the particle-node graph is constructed,and different thread tasks are established based on the already divided regions.When performing particle neighborhood search,different regions are assigned to different threads for search calculation due to the relative independence between regions.This parallel method can effectively reduce the time required for each step of neighborhood particle search,greatly improving the efficiency of the algorithm.This paper validates the performance of the proposed algorithm through experiments in a dam break experiment and compares it with several mainstream search algorithms.The experimental results verify the efficiency of the proposed algorithm,and the parallel experiment validates that the search efficiency of the algorithm is effectively improved in a parallel environment.