| 基于Huffman最优二叉树支持向量机的舱音记录器背景信号识别 |
| |
| 引用本文： | 杨琳,王从庆,缪鹏,姜龙生.基于Huffman最优二叉树支持向量机的舱音记录器背景信号识别[J].宇航学报,2011,32(6). |
| |
| 作者姓名： | 杨琳 王从庆 缪鹏 姜龙生 |
| |
| 作者单位： | 1. 中国民用航空科学技术研究院,北京,100028
2. 南京航空航天大学自动化学院,南京,210016 |
| |
| 摘 要： | 飞行器舱音记录器(CVR)记录的舱音信号,通常是语音声、警告声、开关按钮声和背景噪声等混合而成.目前国内对该类信号的分析和辨别主要是计算机译码后利用人耳进行辨听,存在不易准确分辨各种独立声音信号的缺点.针对舱音信号是一种非平稳性的时频信号,提出了基于多尺度最优小波包基的CVR背景信号特征提取算法,将10种典型信号进行小波包分解,以分解得到的子带能量作为信号初始特征,再根据类间最大距离准则选取最优小波包基,从而确定待识别信号最具有代表性的特征向量,最后基于Huffman最优二叉树支持向量机进行CVR背景信号分类.仿真实验结果表明,该方法的平均识别率为94.62%,可以应用于CVR背景声音信号的自动识别.
|
| 关 键 词： | 舱音记录器 非话语背景声音信号 Huffman最优二叉树支持向量机 信号自动识别 |
Signal Recognition Based on Huffman Optimal Binary Tree SVM for CVR |
| |
| YANG Lin,WANG Cong-qing,MIAO Peng,JIANG Long-sheng.Signal Recognition Based on Huffman Optimal Binary Tree SVM for CVR[J].Journal of Astronautics,2011,32(6). |
| |
| Authors: | YANG Lin WANG Cong-qing MIAO Peng JIANG Long-sheng |
| |
| Institution: | YANG Lin1,WANG Cong-qing2,MIAO Peng2,JIANG Long-sheng2(1.Chinese Academy of Civil Aviation Science and Technology,Beijing 100028,China,2.College of Automatic Control Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China) |
| |
| Abstract: | Signals recorded by CVR(Cockpit Voice Recorder) on aircraft are mixed signal composed by voices,switching knob,alarm sound and noises.So far in domestic,the analysis and identification of these acoustic signals are mainly depended on human audition,difficult to separate independent sounds.Because these signals are non-stationary time and frequency signals,an optimal wavelet packet basis-based algorithm is given by using the arbitrary time-frequency decomposition of wavelet packet transform.Initial character... |
| |
| Keywords: | Cockpit voice recorder Non-voice acoustic signal Huffman optimal binary tree SVM Automatic signal recognition |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
