基于扩充Modelica的复杂系统多层次并行仿真

为提高复杂系统多学科建模仿真效率,弥补现有多领域统一建模仿真语言(Modelica)不支持并行的缺陷,提出了一个面向复杂系统建模仿真的多层次并行仿真框架,从仿真任务级、实体级和模型级3个应用层次上充分发掘仿真系统的并行性.将MPI(Message Passing Interface)/OpenMP中的并行元素加入Modelica语言,并通过所提出的代码映射机制将Modelica文本转换为支持并行编程的高级代码(C++),再利用一定的任务分配机制将高级代码在多层次并行平台(多核集群)上高效执行.针对某典型军事复杂系统优化迭代过程进行仿真实验,结果表明上述多层次并行仿真方法能大幅提高仿真效率.

Multi-layer parallel simulation for complex system based on extended Modelica

To improve multi-disciplinary modeling and simulation(M&S) efficiency for complex systems,and overcome the shortcoming of no parallel mechanisms in current multi-domain M&S languages like Modelica,a multi-layer parallel framework for complex system simulation was proposed.Parallel mechanisms of application systems were captured from three layers including task layer,entity layer and model layer.Modelica was extended by inserting typical parallel elements of message passing interface(MPI) and OpenMP.A code mapping strategy was presented to transfer Modelica texts to parallel-support codes like C+ +.Application systems were simulated in multi-layer parallel platforms like multi-core clusters after task allocating.The iterative process of optimization in a typical military complex system was tested using the proposed means.The simulation results show that the efficiency can be improved evidently through the above framework and methods.