基于△MFCC和KNN的挖掘设备声音识别

摘要

This paper introduces the acoustic signaI processing and the pattern recognition technoIogy to identify the ex-cavation equipment.The main research contents are: ①The frequency domain characteristic parameter extraction aIgorithm of acoustic signaI:MeI-Frequency CepstraI Coefficients (MFCC),the first order difference features of MFCC is △MFCC; ②The KNN is empIoyed as the cIassifier to test the acoustic signaI.TheoreticaI anaIysis and experimentaI resuIts show that the△MFCC recognition rate of the four kinds of excavation equipment acoustic signaIs is 86.69%,which can meet the appIica-tion demands.%为保障电缆不受外力破坏，运用声音信号处理和模式识别技术，对挖掘设备声音信号进行识别。主要研究内容有：①声音信号的倒谱域特征参数分析：MeI频率倒谱系数（MFCC），一阶差分MeI频率倒谱系数（△MFCC）；②采用KNN的识别算法进行测试。理论分析和实验结果表明，对于4种设备声音信号，△MFCC识别率可达86．69％，满足实际应用需求。提出的挖掘设备声音信号处理及识别算法具有通用性，可用于地下管道监测等领域。