摘要
针对降雨时土壤含水量变化导致传统土壤近场散射模型误差较大的问题,提出了基于神经网络的含水土壤近场后向散射模型。该模型将影响潮湿土壤近场散射的多种因素作为自变量,以实测数据为训练样本优化人工神经网络结构,提高了不同含水量土壤后向散射系数预测精度。与实测数据的对比分析表明,小于70°入射角情况下不同含水量土壤后向散射模型精度较高,且具有一定的自主学习能力,可满足毫米波引信探测不同土壤的回波信号仿真要求。
Aiming at the traditional soil near field scattering model error is large when soil water content changing during rainfall,a near-field backscattering model of water soil based on neural network algorithm was proposed.In this model,multiple factors affecting the soil near-field scattering were independent variables,measured data was training sample to optimize network structure,prediction precision of scattering coefficient was improved.Compared with the measured data of the analysis showed that when incidence angle was less than 70°,the model had a higher accuracy and certain ability of autonomic learning,which could meet the echo simulation requirements for millimeter wave(MMW)fuze.
作者
田博
李铁
李伟
候亚丽
TIAN Bo;LI Tie;LI Wei;HOU Yali(Science and Technology on Electromechanical Dynamic Control Laboratory,Xi an 710065,China)
出处
《探测与控制学报》
CSCD
北大核心
2018年第6期23-27,共5页
Journal of Detection & Control
关键词
毫米波引信
地表回波干扰
含水土壤近场散射模型
神经网络
预测精度
MMW fuze simulation
terrain echo jamming
water soil near field scattering model
neural network algorithm
prediction precision