基于Gammatone倒谱系数的直升机声信号识别  

Recognition of Helicopter Acoustic Signal Based on Gammatone Cepstral Coefficients

作　　者：王勇[1] 孟华[2] 陈正武[1] 魏春华 刘垒 WANG Yong;MENG Hua;CHEN Zhengwu;WEI Chunhua;LIU Lei(Key Laboratory of Aerodynamic Noise Control,China Aerodynamics Research and Development Center,Mianyang 621000,China;School of Mathematics,Southwest Jiaotong University,Chengdu 611756,China)

机构地区：[1]中国空气动力研究与发展中心,气动噪声控制重点实验室,四川绵阳621000 [2]西南交通大学数学学院,四川成都611756

出　　处：《湖南大学学报（自然科学版）》2021年第6期74-79,共6页Journal of Hunan University:Natural Sciences

基　　金：国家自然科学基金资助项目(11602290);教育部人文社科青年基金资助项目(18XJC 72040001);试验技术研究资助项目。

摘　　要：直升机辐射的声信号在空气中衰减速度慢、传播距离远,是直升机目标识别的主要依据.受人类听觉系统优异的声音信号识别能力的启发,提出了基于人耳听觉感知特性的Gammatone倒谱系数(Gammatone Cepstral Coefficients,GTCC)特征提取方法用于直升机声信号的识别.通过仿真实验,详细研究了参数设置对识别系统性能的影响,以及识别算法在噪声环境下的鲁棒性、对各种分类器的普遍适用性和相对其他特征提取方法的优越性,对实验结果给出了合理的解释.实验结果表明,所提出的方法能够根据声信号有效地辨识直升机的类别,具有很好的抗噪声干扰能力,有一定的应用前景.不同参数设置对识别性能的影响程度不同,其中窗函数长度、Gammatone倒谱系数特征数量和低频声信号对识别率的影响较大,而高频声信号对目标识别的影响较小.The acoustic signal radiated by the helicopter has a slow attenuation speed and a long propagation distance in the air,which is the main basis for recognition of helicopter target.Inspired by the excellent sound signal recognition ability of the human auditory system,an auditory perceptual feature extraction method based on Gammatone cepstral coefficients(GTCC)is proposed to classify helicopter acoustic signal.Through the simulation experiments,the influence of parameter settings on the recognition performance,the robustness of the proposed method in noisy environments,the universal applicability to various classifiers,and the superiority to other helicopter acoustic signal feature extraction methods were studied in detail.Reasonable explanations of the observed experimental results were also given in this paper.The results show that the proposed method can effectively identify the type of helicopter according to the acoustic signal,and has good anti-interference ability to noise,which has certain application prospects.It is also shown that different parameter settings have different effects on the recognition performance,where the length of window function,the number of Gammatone cepstral coefficient features and the low-frequency acoustic signal have a great influence on the recognition accuracy,while high-frequency acoustic signal has little influence on the recognition.

关 键 词：声信号 识别 直升机 Gammatone滤波器 倒谱系数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] V275.1[自动化与计算机技术—计算机科学与技术]