中文摘要：

针对传统的威胁评估方法存在指标数据冗余、指标权值设置合理性、推理有效性等问题,建立结合网络层次分析法的云推理威胁评估模型,能够合理精简指标,有效优化推理规则。将该模型用于目标识别系统的威胁评估,首先给出威胁评估指标,用网络层次分析法精简指标并得到规范化权值;构建指标云模型,将规范后的指标数据输入前件云发生器,建立推理规则库,引入分级结构简化规则数,运用加权扎德算子实现规则的合成,将合成结果输入后件云发生器得到威胁度云滴,经多次重复操作后,处理数据得到系统威胁度。最后,以实例说明方法的有效性。

英文摘要：

There were some problems in the traditional threat assessment method, such as the index data redundancy, index weight unreasonable and bad reasoning effectiveness. So the cloud reasoning threat assessment model based on analytic network process（ANP） was constructed, which could simplify the reasoning indexes and optimize the inference rules. The model was used to assess the threat degree of different target recognition systems. Firstly, the reasoning indicators were given and simpli- fied by ANP model. Then indexes cloud model were constructed and the standardized indicators data were put into the anteced- ent cloud generators. Then the reasoning rules were constructed which could be reduced by classificaiion structure. The rules were synthetic by weighted Zadeh operator and the synthetic results were put into consequent cloud generator to get the threat degree cloud droplet. After repeating the operation, the system threat degree could be gotten. Finally, the example shows the method is effective and reasonable.