一种基于卷积神经网络的蒸发波导反演方法  

Inversion of evaporation duct based on convolution neural network

作　　者：杨超 王钰鹭 范华磊 靳雨生 YANG Chao;WANG Yulu;FAN Hualei;JIN Yusheng(School of Science,Xi’an University of Posts and Telecommunications,Xi’an 710121,China)

机构地区：[1]西安邮电大学理学院,陕西西安710121

出　　处：《西安邮电大学学报》2023年第3期45-51,共7页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省自然科学基础研究计划项目(2019JQ-200)。

摘　　要：海上蒸发波导的准确反演对提升海上雷达系统的性能有重要的意义。为了提高海洋蒸发波导的反演精度,提出一种基于卷积神经网络的蒸发波导反演模型。构建蒸发波导两参数折射率模型,采用抛物方程方法模拟雷达海杂波传播功率-蒸发波导仿真数据库,使用卷积神经网络模型中的卷积层对海杂波传播功率的特征进行提取与融合,将提取的特征通过全连接层映射到波导参数上,建立了相应的蒸发波导反演模型,反演海洋蒸发波导。仿真结果表明,与其他算法相比,所提算法的反演精度为98%,具有更高的准确性和稳定性。The accurate inversion of the evaporation duct at sea is of great significance for improving the performance of marine radar systems.In order to improve the inversion accuracy of ocean evaporation ducts,a convolutional neural network-based evaporation duct inversion model is proposed.The two-parameter refractive index model of evaporation duct is constructed.The parabolic equation method is adopted to simulate the simulation database of the radar sea clutter wave propagation power-evaporation duct.The characteristics of sea clutter wave propagation power are extracted and fused using the convolutional layer in the convolutional neural network model,and then the extracted characteristics are mapped to the duct parameters through the fully connected layer,and the corresponding evaporation duct inversion model is established to inverse the evaporation duct at sea.The simulation results show that compared with other algorithms,the inversion accuracy of the proposed algorithm is 98%,which has higher accuracy and stability.

关 键 词：蒸发波导 抛物方程 卷积神经网络 雷达海杂波 传播损耗 

分 类 号：TN011[电子电信—物理电子学]