基于关联关系的仿真模型实时智能推荐方法

当全球导航卫星系统(global navigation satellite system,GNSS)分布式仿真环境中共享的模型数量非常多时,检索模型和配置仿真任务将成为一个比较复杂的工程.为提高仿真模型选取和仿真任务配置的效率,设计了一套针对GNSS分布式仿真环境中仿真模型的实时智能推荐方法,方法中首先定义了模型关联关系和接口形状的概念,然后提出了一种条件约束下的频繁模式树(FP-tree)结构,并从理论上分析了该结构在检索任务量方面的减少程度,设计并推导了模型关联关系度的计算方法,以及整套智能推荐方法的运行流程.推荐方法在GNSS分布式仿真环境中进行了仿真验证,仿真结果与传统智能推荐方法做对比分析,分析结果表明,该方法针对仿真模型推荐时运行时间短,推荐结果准确度高,能够实时为用户推荐合适的模型. 

Real-time intelligent recommendation method of a simulation model based on incidence relation

With the availability of a large number of sharing models, model search and task design would be an extremely complex project in the global navigation satellite system (GNSS) -distributed simulation environment (GDSE). For improving the efficiency of model search and task design, a real-time intelligent recommendation method was designed for GDSE. Based on the characteristics of the simulation model, the incidence relation and interface shape of the model were defined in the method and a conditional frequent pattern tree (FP-tree) structure was designed to further improve the retrieval efficiency. The effect of the conditional FP-tree structure was proved theoretically. Then, the calculation method of the model incidence relation degree was proposed and derived based on the Bayesian statistical method. The entire processing of the intelligent recommendation method was designed for implementing it in GDSE. Hence, to check the effect of the real-time intelligent recommendation method, it was implemented in GDSE. Compared with the simulation result of the traditional recommendation method, the model intelligent recommendation method is proved to have a shorter running time and a high accuracy on simulation model recommendation. The computing capability and real-time performance are proved through the simulation. It is demonstrated that the intelligent recommendation method is efficient and flexible for GDSE.