基于EWT-SVM的雨量识别方法  

Rainfall recognition method based on EWT SVM

作　　者：施成龙 行鸿彦[1] 娄华生 Shi Chenglong;Xing Hongyan;Lou Huasheng(School of Electronic and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044)

机构地区：[1]南京信息工程大学电子与信息工程学院,南京210044

出　　处：《气象水文海洋仪器》2024年第1期5-8,共4页Meteorological,Hydrological and Marine Instruments

基　　金：国家重点研发计划(2021YFE0105500);国家自然科学基金(62171228)资助

摘　　要：为了从雨声信号中识别出雨量的大小,提出了一种基于经验小波变换和支持向量机的雨量识别算法。对于采集到的雨声信号先进行去噪,接着对信号进行经验小波变换分解,分解后得到数个经验小波函数分量,然后通过Matlab编程对各个经验小波函数分量进行特征提取,在时域和频域范围内组成评价特征矩阵,最后通过SVM对特征矩阵进行分类识别。通过仿真实验发现,对于同一个信号,经验小波函数相较于经验模态分解有更好的自适应性并且克服了经验模态分解的混叠现象和端点效应。实验结果表明基于经验小波变换和支持向量机的雨量识别方法在雨量识别领域具有良好的效果,研究方法为雨量识别、智能雨量计的发展奠定了良好的基础。In order to identify the magnitude of rainfall from the rain sound signals,a rainfall recognition algorithm based on emprirical wowelet transform and support vector machine is proposed in this paper.The collected rain sound signal is denoised first,then the signal is decomposed by empirical wavelet transform,and several empircal wavelet function components are obtained,and then feature of each empircal wavelet function component is extracted through Matlab,forming an evaluation feature matrix in time and frequency domains.Finally,the feature matrix is classified and identified by SVM.Through simulation experiments,it is found that for the same signal,empirical wavelet transform has better adaptability than empirical mode decomposition and overcomes the aliasing phenomenon and endpoint effect of empirical mode decomposition.The research results show that the empirical wavelet transform and support vector machine method has good performance in the field of rainfall recognition and the research method lays a good foundation for the development of rainfall recognition and intelligent raingauge.

关 键 词：经验小波变换 支持向量机 特征矩阵 雨量识别 

分 类 号：P414.95[天文地球—大气科学及气象学]