摘要: |
利用辐射源目标的频移动态特征进行盲分选时,针对已有的分选方法不能很好解决非合作接收数据非周期、间断、交叠的问题,提出了一种改进的逻辑回归分选算法,将信号分选转化为轨迹分类的问题来处理。首先,借鉴多目标航迹起始模型中的M/N逻辑法的框架,设计新的数据关联门限,以建立目标起始曲线;其次,采用最小二乘拟合方法替代卡尔曼滤波,设计新的外推规则,建立载频变化预测曲线;分析新的接收数据与预测曲线的关联性,不断迭代回归或形成新的分支,最后获得目标分选结果。仿真与实测试验表明,利用逻辑回归分选算法,对不同运动目标搭载的同频辐射源可进行自动分选,4个仿真目标与2个真实目标的分选正确率达到100%。同时,该方法拓展了辐射源目标盲分选中的动态频移特征应用,为辐射源个体识别提供了新的支持。 |
关键词: 辐射源目标分选 多普勒频移 动态可变特征 M/N逻辑法 逻辑回归分选 |
DOI: |
|
基金项目: |
|
Blind Sorting of Emitter Targets Based on Dynamic Characteristics of Frequency Shift |
LE Bo,ZHANG Shuai,WANG Guiliang,HUANG Yuanling,GAN Wenya |
(National Key Laboratory of Science and Technology on Blind Signal Processing,Chengdu 610041,China) |
Abstract: |
This paper introduces an improved logical regression algorithm for the blind target sorting of emitter.While using dynamic characteristics of the emitter target frequency shift for blind sorting,the regular sorting algorithms may suffer from non-periodic,intermittent and overlapping issues of the non-cooperative received data.First,a similar working frame of the M/N logic method commonly used in the multi-target tracking initiation is adopted to set up a new data association threshold and establish the target initiation curve.Instead of the Kalman filter,the least square fitting method is used to set up a new extrapolation rule,thus creating a prediction curve of the carrier frequency shift.Through an iterative curve fitting process based on the correlation between the new received data and the prediction curve,the final target sorting result is obtained.Simulation and test results show that the emitters with the same frequency carried by different moving targets can be automatically sorted by using this logic regression sorting algorithm.One hundred percent of the sorting accuracy in four simulation targets and two real targets is achieved.Meanwhile,this method extends the application of dynamic characteristics of frequency shifts in the blind target sorting,and provides a new support for the individual emitter recognition. |
Key words: emitter target sorting Doppler shift dynamic variable feature M/N logistic method logistic regression sorting |