数据缺失状态下目标威胁评估的AR(p)动态突变排序法

AR(p) dynamic catastrophe ranking method of target threat assessment under the loss of data

针对目标威胁评估过程中观测数据缺失、难以综合考虑来袭目标的动态威胁态势以及已有评估方法大多需要确定各指标权重而过多依赖专家经验的问题,将自回归(auto regressive,AR(p))预测模型、时间序列赋权及突变理论相结合,提出了目标威胁评估的AR(p)动态突变排序法。该方法通过AR(p)模型预测时间序列上的缺失目标数据,为综合各时刻数据进行威胁评估提供了基础数据信息;采用泊松分布逆形式对时间序列进行赋权,全面考虑了当前时刻和之前时刻的目标关联信息;在突变理论基础上抽象出了突变排序的思想,将目标威胁评估指标体系中下层指标对上层指标的作用归因为控制变量对状态变量的作用,并采用归一化公式对作用程度进行求解,以此方式由底层往上逐层推进最终得到目标威胁评估值。实例分析结果表明,AR(p)动态突变排序法适用于目标数据缺失时的威胁评估,更贴合实战,且不需确定各指标权重,计算过程简洁高效,可操作性强,由于综合考虑了目标的动态威胁态势,评估结果更加合理,具有一定的实用价值。

For the problems of the loss of observation data, the hardness on overall consideration of dynamic threat situation of the coming target, the need of confirming index weights of most existing methods and too much rely on expertise, the AR（p） dynamic catastrophe ranking method of target threat assessment is proposed by combining the AR（p） predicting model, time series weight and catastrophe theory. The AR（p） model is used to predict the missing target data on time series , which provides the basic data for threat assessment using all times, data. Time series weight is gained by the inverse form of the Poisson distribution, and the target as-sociating information of the current time and the former time is comprehensively considered. The core thoughts of catastrophe ranking is abstracted based on the catastrophe theory, and the function of sub layer indexes on its upper layer indexes in the target threat assessment index system is attributed to the function of control variable on state variable. The degree of this function is solved by the normalization formulae, and the target threat as-sessment value is acquired through the bottom layer to the top layer according to this mode. The experiment result indicates that the AR（p） dynamic catastrophe ranking method can be used for target threat assessment under the loss of target data, which is more close to the actual combat. It is unnecessary to compute the weights of all indexes, which is concise, efficient, and easy to implement. In addition, the assessment result is more reasonable and has a certain ap-plication value due to the comprehensive consideration of the dynamic threat situation of targets.