Abstract: When dealing with incoming Unmanned Aerial Vehicles (UAVs), it is crucial to quickly and accurately detect and identify the formation of enemy UAVs in order to analyze and judge their combat intentions and formulate effective countermeasures. An adaptive threshold method was established to construct a multi-scale complex network of the unmanned cluster formation, and the combination of eigenvalues corresponding to the adjacency matrix of these complex networks was selected to form a shape signature. By introducing the Hellinger distance to measure the difference between the shape signatures of the formation to be identified and the standard formation, the identification result is obtained. Simulation results show that, compared with the method of obtaining multi-scale complex networks with hard thresholds, the proposed algorithm has better adaptability and robustness, and has a higher recognition rate even when the target information is severely contaminated. Moreover, the algorithm has fewer parameters and lower time complexity.

Key words: multi-scale complex network, formation recognition, spectral graph theory, Hellinger distance, Unmanned Aerial Vehicle (UAV)

摘要： 在面对来袭无人机（UAV）时，快速准确地检测识别出敌方UAV的编队队形，对于分析判断敌方的作战意图、制定有效反制措施至关重要。因此，建立自适应阈值方法将无人集群队形构建为多尺度复杂网络，选择这些复杂网络对应的邻接矩阵的特征值组合，形成形状签名。通过引入Hellinger距离度量待识别队形与标准队形的形状签名间的差异性，从而得到识别结果。仿真结果表明，与硬阈值得到多尺度复杂网络的方法相比，所提算法具有较好的适应性和鲁棒性，在目标信息受污染较严重时，也具有较高的识别率，且算法具有较少的参数和较低的时间复杂度。

关键词: 多尺度复杂网络, 队形识别, 谱图理论, Hellinger距离, 无人机