基于极限学习机的低频雷达自适应测高技术

Adaptive Altitude Measurement Based on Extreme Learning Machine in VHF Array Radar

在低频段雷达低仰角测高中,由于直达波和相干多径信号都在天线波束的主瓣之内,导致测高精度低。提出了一种基于极限学习机的低频段雷达测高技术,利用合作目标的回波数据和真实仰角对极限学习机进行训练,对非合作目标的仰角测量利用训练的极限学习机对目标仰角进行预测。相比阵列信号处理技术,该算法不需要雷达反射面的任何参数,具有良好的鲁棒性和泛化性。实测数据处理结果证实了该算法的可行性。

The multi-path signal and the direct signal,lying within a beam-width of the receiving antenna,are highly correlated,which seriously affect the performance of low-angle altitude measurement for very high frequency(VHF)radar.A novel adaptive altitude measurement based on Extreme Learning Machine(ELM)was therefore proposed.ELM was trained by the echo signal and elevation information of cooperative targets;and the elevation angle measurement on non-cooperative targets was estimated by the trained ELM.Compared to the traditional array signal processing techniques,the proposed method required no parameters of reflection region,featured with the robustness and good generalization performance.The real data processing results demonstrated the validity and feasibility of the proposed method.