基于CUDA技术的振动筛离散颗粒运动仿真

摘要

在以往筛分过程中,对筛分效率和透筛率的计算均采用传统的和基于概率统计的方法.本文基于离散元理论,采用CUDA技术,利用GPU高并行度、多线程、强大计算能力和极高存储器带宽的特点,模拟大规模颗粒在筛分过程中的运动情况.在此基础上,寻找对离散元数值计算的可行方法,并改进程序,使其能高效地模拟颗粒运动,为相关结构的改造提供依据.%In the past,during the process of screening,the screening efficiency and sieving rate calculations all took traditional way that based on the probability and statistics method.Using CUDA technology which has characteristics of high degree of parallelism,multithreading,powerful computing capability and high memory bandwidth and discrete element method,it is convenient to simulate the movement of massive particles in the process of screening.On the basis of previous works,the feasible way of discrete element numerical calculation methods can be found and efficient simulation of particle movement could be designed in order to provide a basis for the future transformation.