基于HLA的Petri网分布式仿真中的数据过滤方法

在应用Petri网对大规模复杂系统建模和仿真时,Petri网模型可能出现组合爆炸的现象,从而占用大量的计算资源。考虑到Petri网的并发特性,可将Petri网模型分解为若干子网,进行并行或分布式仿真。在基于HLA的Petri网分布式仿真中,各联邦成员之间需要进行大量的数据交换,从而实现令牌(token)在子网之间的流动。因此,减少冗余数据的传输是提高分布式仿真性能的一种有效途径。对基于HLA的Petri网分布式仿真中的四种数据过滤方法进行了分析和研究,并通过仿真实验比较了它们去除冗余数据的有效性。

Data Filtering in HLA-Based Distributed Simulation of Petri Nets

Petri net is a widely used mathematical tool for complex systems modeling and simulation. The combinational explosion of Petri nets may occur and excessive computation resources are required to run the simulation especially for large-scale complex systems. To solve this problem, the Petri net models can be decomposed into several subnets based on the concurrency property of the Petri nets, and the simulation can be executed in parallel or distributed way. In the HLA-based distributed simulation of Petri nets, federates must exchange a great deal of data to describe the token flow among the subnets when the state changes in the simulated systems. So reducing the redundant transmission of data is an effective way to improve the performance of the distributed simulation. Four data filtering methods were discussed and implemented in HLA-based distributed simulation of Petri nets. The experimental results were represented to compare the efficiencies of these methods.