基于支持向量机的雷达地物回波识别研究
投稿时间: 2014-08-15  最后修改时间: 2014-09-16  点此下载全文
引用本文:
摘要点击次数: 408
全文下载次数: 
作者单位E-mail
魏鸣 南京信息工程大学 mingwei@nuist.edu.cn 
基金项目:国家重点基础研究发展计划(973计划);国家高技术研究发展计划(863计划)
中文摘要:多普勒天气雷达探测过程中的非气象因子会显著影响雷达资料的定量化应用,在应用雷达基数据之前必须对雷达资料进行抑制地物杂波、去距离折叠和退速度模糊等质量控制。本文在现有的自动识别地物回波方法的基础上,提出了基于支持向量机(Support Vector Machine,SVM)识别雷达地物杂波的方法,对安庆和常州两地2013年6月—8月的CINRAD/SA雷达观测资料进行雷达地物回波识别,并将其与运用人工神经网络(Artificial Neural Networks,ANNs)识别的结果进行对比,表明支持向量机方法能够取得更好的效果地物回波的准确识别率明显高于神经网络方法的识别结果。在训练样本较少时,支持向量机方法的准确识别优势尤为突出。
中文关键词:地物回波  神经网络  支持向量机
 
The Ground Clutter Identification Based on SVM Method with the Doppler weather radar data
Abstract:Non-meteorological radar echo has a serious impact on the quantitative application of Doppler radar data, imperative quality control methods such as ground clutter elimination, range folding elimination, velocity dealiasing must be adopted before using radar data. On the basis of existed automatic algorithm for ground clutter detection, this paper proposes a SVM (Support Vector Machine) based method to precisely identify the ground clutter echo and precipitous echo, and identifiable recognition accuracy of ground clutter and precipitous echo with SVM and ANNs(Artificial Neural Networks) is compared using the data observed by Doppler radars in Changzhou and Anqing during June to August 2013. The results shows that SVM based method does better in the radar echo identification, especially in the identification of ground clutter echo, compared to the ANNS based method. Especially on the condition of small sample, SVM based method gets a much better results.
keywords:ground clutter  artificial neural networks  support vector machine
查看全文  查看/发表评论  下载PDF阅读器