基于ECMDA-EN的空战目标威胁评估

王昱; 谭丞志; 梁宵; 范子琳; 任田君

doi:10.3969/j.issn.2095-1248.2023.05.006

沈阳航空航天大学学报 ›› 2023, Vol. 40 ›› Issue (5) : 38 -49. DOI: 10.3969/j.issn.2095-1248.2023.05.006

信息科学与工程

基于ECMDA-EN的空战目标威胁评估

作者信息 +

Aerial combat target threat assessment based on ECMDA-EN

Author information +

文章历史 +

PDF (1288K)

摘要

强对抗空战环境下获取的态势信息常伴有错误、缺失等缺陷数据。由于模型和参数相对固定且缺少有针对性的信息修正环节，现有的目标威胁评估方法极易导致相关时刻威胁评估结果失真。提出了一种基于证据修正及模型动态调整机制的证据网络（Evidence correction and model dynamic adjustment evidential network，ECMDA-EN）的空战目标威胁评估方法。首先，基于证据信息的突变型、连续型和确定型类型划分，结合预判的信息缺陷程度设计4种证据修正方式。在此基础上，针对连续型证据信息，通过汇总不同作战意图下的空战对抗轨迹构造样本数据，提出基于LSTM神经网络轨迹预测模型的缺陷数据修正方法。最后，针对证据节点删除的情况设计了网络节点权重以及模型的动态调整方案。仿真实例表明：在连续推理的过程中，ECMDA-EN方法依靠证据修正方式的合理切换，具备对缺陷信息的有效处理以及对模型的自适应调整能力，解决了传统威胁评估方法过于依赖信息可靠性的问题，能够持续给出合理的威胁评估结果。

Abstract

The situation information obtained in the environment of strong confrontation air combat is often accompanied by error，missing and other defective data. Due to the relatively fixed model and parameters and the lack of targeted information correction links，the existing air combat target threat assessment methods are prone to lead to the distortion of the threat assessment results at the relevant time. A target threat assessment method of evidence correction and model dynamic adjustment evidential network （ECMDA-EN） was proposed. Firstly，based on the classification of mutant，continuous and definite types of evidence information，four kinds of evidence correction methods were designed combined with the degree of predicted information defect. On this basis，the sample data was constructed by summarizing the air combat confrontation trajectory under different combat intentions，and the defect data correction method based on a LSTM neural network trajectory prediction model was proposed. Finally，a dynamic adjustment scheme of network node weights and model was designed according to the deletion of evidence nodes. Simulation results show that the proposed method solves the problem that the traditional threat assessment method rely too much on the reliability of information by effectively processing the defect information and automatically adjusting model，and can continuously provide reasonable threat assessment results.