一种雷声实时识别方法  

A Method for Real-time Recognition of Thunder

作　　者：卫克晶[1] 李浩然[1] 孙学金[1] 张传亮[1] 

机构地区：[1]解放军理工大学气象海洋学院,江苏南京211101

出　　处：《实验室研究与探索》2015年第3期27-30,共4页Research and Exploration In Laboratory

基　　金：国家自然科学基金资助项目(41205125)

摘　　要：针对传统雷声识别方法识别速度慢、稳定性不佳的问题,提出一种基于端点检测和特征提取的雷声快速识别方法。通过端点检测提取出异常声音,建立基于频谱信息的特征向量,求得异常声音信号的特征向量与实验得出的雷声标准特征向量之间的相关系数,用阈值法判别异常声音信号是否为雷声。多种干扰源和信噪比测试结果表明,该方法的识别速度快,识别效率高。Due to the low recognition speed and poor stability of traditional method in thunder recognition, in this paper, a method for real-time recognition of thunder was put forward based on endpoint detection and feature extraction. Firstly, the abnormal acoustic signal was extracted by endpoint detection. Then, the eigenvector based on frequency spectrum was established and the related coefficient between the eigenvectors of the abnormal acoustic signal and standard thunder acoustic signal that was measured by experiment was obtained. Finally, the threshold method was introduced to identify whether the abnormal acoustic signal was thunder or not. The tests of interfering and noise superpositions showed that the method for identifying thunder is fast and highly efficient.

关 键 词：雷声识别 端点检测 特征提取 相关系数 信噪比 

分 类 号：P427.42[天文地球—大气科学及气象学]