作者：邹瑛珂¹, 李祖明², 刘晓宏³, 贾云飞¹

作者单位：1. 南京理工大学机械工程学院，江苏 南京 210094;
2. 南京工程学院电力学院，江苏 南京 211167;
3. 盐城供电公司，江苏 盐城 224000

关键词：特征提取;无人机;变分模态分解;梅尔倒谱系数;GammaTone倒谱系数;随机森林

摘要：

为解决传统声信号特征在环境中对含有大风、街道常见人造声音干扰的无人机噪声信号识别率较低的问题，该文提出一种基于VMD分解的MFCC+GFCC无人机噪音混合特征提取方法。首先，对目标的声音信号进行VMD分解，获得各IMF信号和原始信号的能量之比；然后，利用已获得的信号进行MFCC/GFCC系数提取，并获得二者的一阶差分系数；最后，使用随机森林分类算法对信号进行分类，从而实现对无人机噪声信号的正确识别。结果表明：识别准确率比单MFCC/GFCC等传统特征提取方法在含噪或纯净无人机噪声条件下识别率提升4%以上。

Hybrid feature extraction method of MFCC + GFCC UAV noise based on VMD decomposition
ZOU Yingke¹, LI Zuming², LIU Xiaohong³, JIA Yunfei¹
1. College of Mechanical Engineering, Nanjing University of Technology, Nanjing 210094, China;
2. School of Electric Power Engineering, Nanjing Institute of Technology, Nanjing 211167, China;
3. Yancheng Power Supply Company, Yancheng 224000, China
Abstract: In order to solve the problem that the recognition rate of traditional acoustic signal features is low for UAV noise signals with wind and common artificial noise interference in streets, a method of extracting mixed noise features of MFCC + GFCC UAV based on VMD decomposition is proposed. Firstly, the voice signal of the target is decomposed by VMD to obtain the energy ratio of each IMF signal and the original signal; then the MFCC / GFCC coefficients of the obtained signal are extracted and the first-order difference coefficients of the two are obtained; finally, the random forest classification algorithm is used to classify the signal, so as to realize the correct recognition of UAV noise signal. The recognition accuracy is improved by more than 4% compared with the traditional feature extraction methods such as single MFCC / GFCC under the condition of noisy or pure UAV noise.
Keywords: feature extraction;UAV;variational mode decomposition;MEL cepstrum coefficient;GammaTone cepstrum coefficient;random forest
2021, 47(11):141-146  收稿日期: 2021-04-20;收到修改稿日期: 2021-05-28
基金项目: 国网江苏省电力有限公司科技项目（J2021041）
作者简介: 邹瑛珂（1996-），男，重庆市人，硕士研究生，专业方向为信号处理与模式识别
参考文献
[1] 胡焱, 徐志强, 刘文劲, 等. 基于RetinaNet的低小慢无人机目标识别[J]. 现代计算机, 2021, 4(5): 66-70, 74
[2] 金恒康, 张一闻, 王耀杰. 无人机飞行声特征与提取方法比较[J]. 现代电子技术, 2019, 42(22): 103-107, 112
[3] 丘恺彬, 李建良. 无人机识别的音频特征提取方法[J]. 噪声与振动控制, 2018, 38(2): 188-192
[4] JEON S, SHIN J W, LEE Y J, et al. Empirical study of drone sound detection in real-life environment with deep neural networks[J]. // 25th European Signal Processing Conference (EUSIPCO), 2017.
[5] JEGADEESHWARAN R, SUGUMARAN V, SOMAN K P. Vibration based fault diagnosis of a hydraulic brake system using variational mode decomposition (VMD)[J]. Structural Durability & Health Monitoring, 2014, 10(1): 81-97
[6] 句海洋, 王新华, 赵以振. 改进VMD钢质管道损伤信号提取算法研究[J]. 中国测试, 2020, 46(5): 100-107
[7] 屠彬彬, 于凤芹. 基于EMD的改进MFCC的语音情感识别[J]. 计算机工程与应用, 2012, 48(18): 119-122
[8] 周萍, 沈昊, 郑凯鹏. 基于MFCC与GFCC混合特征参数的说话人识别[J]. 应用科学学报, 2019, 37(1): 24-32
[9] GAO P, ZHAO J, WANG G, et al. Real time ECG characteristic point detection with randomly selected signal pair difference (RSSPD) feature and random forest classifier[C]//2016 38th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2016.
[10] 魏静明, 李应. 基于纹理特征与随机森林的生态声音识别[J]. 计算机应用与软件, 2015, 32(3): 162-166
[11] ZHAO Y N. Prediction and assessment of air pollution in shandong province based on CART decision tree and radial basis function neural network[J]. Statistics and Applications, 2019, 8(5): 725-733