机场场面多传感器多轴向感知信号的融合方法

A Multi-axis Signal Fusion Approach Using Multiple Sensors of the Aerodrome Surface

为实现对机场场面航空器/车辆的识别,提出基于D-S证据理论的多轴向传感器二级融合方法,对多个磁阻传感器获取的非协作目标感知特征进行融合。通过最大隶属度原则确定单轴向基本概率分配函数,采用基于证据间相似系数的一级融合方法确定单传感器的3个轴向合成概率分配函数;考虑到传感器网络中的各传感器的感知磁信号的差异性,通过地磁信号能量值的方法分配各证据权重,利用加权综合法进行二级融合。实验结果显示,不考虑3轴向之间相关性及证据间信号能量权重的融合方法的总体识别率仅为63.1%;仅考虑3轴向之间相关性的融合方法的总体识别率为65.6%;仅考虑证据间信号能量权重的融合方法的总体识别率为69.6%;而基于证据间相似系数和信号能量值权重二级数据融合方法的总体识别率高达81.1%。

To recognize the aircrafts or vehicles operating on the airport surface,a two-stage multi-axis data fusion algorithm based on Dempster-Shafer（D-S）evidence theory is proposed,so as to integrate the features of non-cooperative targets derived from multiple anisotropic magnetoresistive sensors.For each sensor,a basic probability function for every single axis based on the principle of maximum membership is assigned,and then the first-step fusion through constructing the synthesis probability distribution function of the three-axis signals is completed through the similarity coefficients between evidences.Because the magnetic signal of each sensor differs with each other,the corresponding energy measure of geomagnetic signals is used to assign the weight of each evidence,and finally a weighted synthesis is applied to complete the second-step of the data fusion.Simulation results indicate that the recognition rate is only 63.1% when neither the relevance among three-axis signals nor the weight of evidences is considered.The recognition rate is 65.6% when relevance among three-axial signals is considered only;and it is 69.6% when signal energy weight among evidences is considered only.The recognition rate increases to 81.1% when the proposed two-stage data fusion algorithm is applied and both similarity coefficients of evidences and signal weight are taken into account.