| 本文已被：浏览 1190次 下载 959次 |
码上扫一扫！ |
| 基于改进EMD-TEO倒谱距离的生猪音频信号端点检测 |
|
吴亚文1,邵睿1,李淼2,张锋1,陶浩兵1,辜丽川1,焦俊1*
|
|
|
| (1.安徽农业大学 信息与计算机学院, 合肥 230036;2.中国科学院 智能机械研究所, 合肥 230031) |
|
| 摘要: |
| 针对生猪音频信号识别中传统端点检测方法存在的抗噪能力差，准确率低等问题，将一种改进EMD-TEO(经验模态分解的Teager能量算子)倒谱距离的端点检测方法应用到生猪音频信号的端点检测，对生猪音频信号的2个端点检测特征参数(Teager能量参数和倒谱距离参数)进行研究。仿真试验结果表明:1)与EMD-TEO和倒谱距离端点检测法相比，当高斯白噪声的信噪比(Signal-noise ratio，SNR)为-5 dB时，改进后的算法用于生猪音频信号的端点检测准确率达到84.725%，分别高出前2种端点检测算法9.857%和16.403%，SNR为10 dB时，仍优于其他2种算法；2)当饲养场背景噪声的SNR为10 dB时，改进后的用于生猪音频信号的端点检测准确率达到90.293%，分别高出EMD-TEO和倒谱距离端点检测算法4.972%和7.932%；3)改进后的算法对不同类型的生猪音频信号均呈现出很好的端点检测效果。改进EMD-TEO倒谱距离的端点检测算法在低SNR的环境下，对生猪音频信号的端点检测有较高正确率，且具有一定的鲁棒性。 |
| 关键词: 生猪音频信号 经验模态分解 Teager能量算子 倒谱距离 端点检测 |
| DOI：10.11841/j.issn.1007-4333.2021.04.10 |
| 投稿时间：2020-07-17 |
| 基金项目:国家自然科学基金项目(31671589)；安徽省科技重大攻关项目(16030701092)；安徽省重点研究与开发计划项目(1804a07020130)；安徽省科技重大专项(201903a06020009) |
|
| Endpoint detection of live pig audio signal based on improved EMD-TEO cepstrum distance |
|
WU Yawen1,SHAO Rui1,LI Miao2,ZHANG Feng1,TAO Haobing1,GU Lichuan1,JIAO Jun1*
|
| (1.Division of Information and Computer, Anhui Agricultural University, Hefei 230036, China;2.Institute of lntelligent Machines, Chinese Academy of Sciences, Hefei 230031, China) |
| Abstract: |
| Aiming at the problem of poor antinoise ability and low accuracy in traditional endpoint detection method in pigs audio signal recognition, an improved EMD-TEO(Teager energy operator of empirical mode decomposition)cepstrum distance of endpoint detection method was applied to the pig audio signal endpoint detection. The characteristic parameters of two pigs audio signal endpoint detection were studied. The simulation results show that: 1)Compared with EMD-TEO and cepstrum distance endpoint detection method, when the signal-to-noise ratio(SNR)of Gaussian white noise is higher than that of EMD-TEO and cepstrum distance endpoint detection method, the signal-to-noise ratio(SNR)of Gaussian white noise is higher than that of emd-teo and cepstrum distance endpoint detection. When SNR is -5 dB, the accuracy rate of the improved algorithm is 84. 725%, which is 9. 857% and 16. 403% higher than the first two endpoint detection algorithms, respectively. When SNR is 10 dB, it is still better than the other two algorithms; 2)When the SNR of the background noise is 10 dB, the accuracy rate of the improved endpoint detection for pig audio signal reaches 90. 293%, which is respectively 4. 972% and 7. 932% higher than EMD-TEO and cepstrum distance endpoint detection algorithm respectively; 3)The improved algorithm has good endpoint detection effect for different types of pig audio signals. In conclusion, the improved EMD-TEO cepstrum distance endpoint detection algorithm can achieve high accuracy and robustness in the low SNR environment. |
| Key words: pig audio signal EMD TEO cepstrum distance endpoint detection |
