海底混响非瑞利特性研究及神经网络应用  

Research on non-rayleigh characteristics of seafloor reverberation and its application of neural network

作　　者：刘罡 杨云川[1] LIU Gang;YANG Yun-chuan(The 705 Research Institute of CSIC,Xi′an 710077,China)

机构地区：[1]中国船舶重工集团公司第705研究所,陕西 西安 710077

出　　处：《舰船科学技术》2020年第5期123-126,共4页Ship Science and Technology

摘　　要：基于单元散射理论介绍了瑞利分布模型和K分布模型,通过计算混响偏度和峰度判断出海底混响偏离瑞利分布模型,并利用CW信号、LFM信号的试验混响数据进行阵元域、波束域上的PDF曲线拟合。结果表明,海底混响的统计特性更趋向于K分布模型。利用BP网络方法和海底混响、点目标仿真信号的PDF特性进行了目标识别验证,其正确识别率达到了92%以上,且计算量大大降低。In this paper,based on the element scattering theory,the Rayleigh distribution model and the K-distribution model are introduced.By calculating the skewness and kurtosis of the reverberation,it is judged that the seafloor reverberation deviates from the Rayleigh distribution model,and the PDF curve fitting on the array element domain and the beam domain is performed by using the experimental reverberation data of the CW signal and the LFM signal.This result shows that the statistical characteristics of seafloor reverberation tend to be K-distribution model.Finally,by using BP neural network method and PDF characteristics of simulation data of seafloor reverberation and point target signals,the target recognition is verified,and the final correct recognition rate is over 92%,and the calculation amount is greatly reduced.

关 键 词：海底混响 K分布 统计特性 曲线拟合 神经网络 

分 类 号：TN011[电子电信—物理电子学]