证据理论与模糊函数相结合的弹道目标识别

Using evidence theory and fuzzy function to identify ballistic target

以稳健性为原则,选定物理意义明确、提取难度较低、类别可分性强、特征值起伏度小的若干种特征进行弹道中段目标群的识别研究。通过建立各类特征的模糊隶属度函数,完成相应不同证据体的基本概率赋值。再利用重新定义的证据理论冲突系数,对冲突或较低可信度证据进行修正。最后,尝试将模糊函数与改进证据理论相结合的算法应用于弹道中段的目标识别,仿真结果表明该无监督模式识别方法能够可靠地识别出真实弹头,也更具有良好的泛化能力。

Several kinds of characteristics are adopted to identify middle ballistic target group by characteristics’stability principle; for these characteristics, physical meaning is clear, extraction difficulty is low, separability is strong and value fluctuation is small. Firstly, according to characteristics’variation law, fuzzy recognition model of real warhead is established and basic probability assignment of different evidences is completed. Then, the conflict coefficient is redefined by using the complementarity between classical conflict coefficient and the distance function in describing evidence conflict, and conflict or lower confidence evidence is revised as well. Lastly, the combination of fuzzy function and improved evidence theory is applied to middle ballistic target recognition by setting simulation conditions. Simulation results show that the proposed method can identify the real warhead successfully.