航空学报 > 2021, Vol. 42 Issue (12): 324768-324768   doi: 10.7527/S1000-6893.2020.24768

基于云模型和改进D-S证据理论的目标识别决策方法

尹东亮1, 黄晓颖1, 吴艳杰1, 何有宸2, 谢经伟3   

  1. 1. 海军工程大学 作战运筹与规划系, 武汉 430033;
    2. 中国人民解放军 91951部队, 烟台 265100;
    3. 海军工程大学 职业教育中心, 武汉 430033
  • 收稿日期:2020-09-21 修回日期:2020-11-03 发布日期:2020-12-03
  • 通讯作者: 黄晓颖 E-mail:hjgchxy@163.com
  • 基金资助:
    国家自然科学基金(71501183)

Target recognition decision method based on cloud model and improved D-S evidence theory

YIN Dongliang1, HUANG Xiaoying1, WU Yanjie1, HE Youchen2, XIE Jingwei3   

  1. 1. Department of Operations Research and Planning, Naval University of Engineering, Wuhan 430033, China;
    2. No. 91951 PLA Troop, Yantai 265100, China;
    3. Vocational Education Center, Naval University of Engineering, Wuhan 430033, China
  • Received:2020-09-21 Revised:2020-11-03 Published:2020-12-03
  • Supported by:
    National Natural Science Foundation of China (71501183)

摘要: 在目标识别决策系统中,多探测器多源信息融合的模糊性和不确定性以及各探测周期所得信息的冲突互斥会造成目标识别决策不精准。为解决这一问题,提出基于云模型和改进D-S (Dempster-Shafer)证据理论的目标识别决策方法。首先,将目标识别准确性这一语言评价值划分为不同评价区间等级,以不同评价等级标准云为参照将各探测器各探测周期所得信息转化为云决策矩阵,得出各周期各等级隶属度,进而构建出基本概率分配函数(mass函数);其次,基于证据理论引入冲突度、差异度、离散度3类衡量冲突大小的参数,定义了一种新的证据冲突参数,同时改进证据冲突融合算法,对各探测器各周期证据体进行修正并融合;再次,结合各探测器权重加权得出各目标综合识别决策的mass函数对目标进行决策;最后,结合算例,验证该方法的适用性,并与其他方法相对比验证了本文方法的优越性。

关键词: 目标识别, 多源信息融合, 云模型, 证据理论, 证据冲突

Abstract: In the decision-making system for target recognition, the ambiguity and uncertainty of multi-detector and multi-source information fusion and the conflict and mutual exclusion of the information obtained in each detection cycle will cause inaccurate recognition of the target. To solve this problem, a decision-making method for target recognition based on the cloud model and improved D-S (Dempster-Shafer) evidence theory is proposed. First, the linguistic evaluation value of target recognition accuracy is divided into different evaluation interval levels, and the information obtained from each detection period of each detector is converted into a cloud decision matrix based on the standard cloud of different evaluation levels to obtain the membership degree of each level of each period, and then a basic probability distribution function (mass function) is constructed. Second, based on the evidence theory, three types of parameters, namely conflict degree, difference degree, and dispersion degree, are introduced to measure the magnitude of conflict. A new evidence conflict parameter is defined, and the evidence conflict fusion algorithm is improved to revise and combine the evidence body of each detector. Thirdly, the mass function of the comprehensive decision making for each target recognition is obtained by combining the weight of each detector to make a decision on the target. Finally, a numerical example is used to verify the applicability of this method, and a comparison with other methods demonstrates the superiority of this method.

Key words: target recognition, multi-source information fusion, cloud model, evidence theory, evidence conflict

中图分类号: