The credibility evaluation method for a single simulation model is relatively mature at present. While there is still a great need to provide an effective credibility evaluation method for complex simulation systems, which is the basic premise to assist the design, test, evaluation and optimization of a complex research object based on constructing its simulation system. In this paper a credibility evaluation method for complex simulation systems based on the interaction analysis of their internal data is proposed. A directed weighted interaction network describing the interaction among models in a complex simulation system can be constructed. Further, based on the credibility of each simulation model and the network structure, we can quantitatively analyze their impacts on the overall credibility of the system. In order to verify the effectiveness of the method, we constructed a validation system based on the network structure of a multi-agent cooperative simulation system with known credibility. The evaluation result comparison showed that our method can provide a feasible way for the relatively accurate credibility analysis of complex simulation systems. Finally, our verified method was applied to evaluate the credibility of the complex hardware-in-the-loop simulation system composed with over 100 components, which provided a quantitative way instead of traditional qualitative evaluation by experts.

目前，单个仿真模型的可信度评估方法已经比较成熟。尽管仍然迫切需要为复杂的仿真系统提供一种有效的可信度评估方法，这是在构建复杂的仿真系统的基础上协助设计，测试，评估和优化复杂研究对象的基本前提。提出了一种基于内部数据交互分析的复杂仿真系统可信度评估方法。可以构建一个定向加权交互网络，该网络描述复杂仿真系统中模型之间的交互。此外，基于每个仿真模型的信誉和网络结构，我们可以定量分析它们对系统总体信誉的影响。为了验证该方法的有效性，我们基于具有已知信誉的多智能体协同仿真系统的网络结构，构建了一个验证系统。评估结果的比较表明，该方法可以为复杂仿真系统的相对准确的可信度分析提供一种可行的方法。最后，将我们经过验证的方法用于评估由100多个组件组成的复杂的硬件在环仿真系统的可信度，这为定量分析提供了一种定量方法，而不是传统的定性评估方法。评估结果的比较表明，该方法可以为复杂仿真系统的相对准确的可信度分析提供一种可行的方法。最后，将我们经过验证的方法用于评估由100多个组件组成的复杂的硬件在环仿真系统的可信度，这为定量分析提供了一种定量方法，而不是传统的定性评估方法。评估结果的比较表明，该方法可以为复杂仿真系统的相对准确的可信度分析提供一种可行的方法。最后，将我们经过验证的方法用于评估由100多个组件组成的复杂的硬件在环仿真系统的可信度，这为定量分析提供了一种定量方法，而不是传统的定性评估方法。