车站运行列车异音检测方法  

ABNORMAL SOUND DETECTION METHOD OF TRAINS RUNNING AT STATIONS

作　　者：张宏睿 马秀荣 单云龙 Zhang Hongrui;Ma Xiurong;Shan Yunlong(School of Electrical and Electronic Engineering,Tianjin University of Technology,Tianjin 300384,China)

机构地区：[1]天津理工大学电气电子工程学院

出　　处：《计算机应用与软件》2019年第8期130-137,171,共9页Computer Applications and Software

基　　金：天津市科技创新专项基金项目(10FDZDGX00400)

摘　　要：针对运行列车异音检测的高实时性和高准确率要求,提出一种改进的最小值控制递归平均噪声估计算法(minima controlled recursive averaging,MCRA)和一种以改进能熵比值为特征值的异音检测算法。根据无法提取纯净行车音频和列车运行环境噪声变化大的特点,改进MCRA算法中最优平滑因子及功率谱最小值跟踪算法,有效解决MCRA算法中存在的噪声估计延时问题和噪声功率谱估计不准确问题。针对异常类型较多的特点,采用改进的能熵比检测算法,有效识别四类异常情况。实验结果表明,结合上述两种方法能够有效确定异常车厢和异常行驶类型,准确率达91%。To meet the requirements of high real-time and high accuracy of abnormal sound detection for running trains,this paper proposed an improved minima controlled recursive averaging(MCRA) algorithm and an abnormal acoustics detection algorithm with improved energy-entropy ratio as eigenvalue.According to the fact that it is impossible to extract the pure driving audio and the large variation of noise in train operation environment,we improved the optimal smoothing factor and power spectrum minima tracking algorithm in MCRA algorithm.The problems of noise estimation delay and inaccurate noise power spectrum estimation in MCRA algorithm were effectively solved.In addition,according to the characteristics of various types of anomalies,the improved energy-entropy ratio detection algorithm was used to effectively identify various anomalies.The experimental results show that the combination of the two algorithms can effectively determine the abnormal carriage and the abnormal driving type,and the accuracy is 91%.

关 键 词：噪声估计 最小值控制递归平均算法 谱减法 异常声音 能熵比检测法 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术] TN912.16[电子电信—通信与信息系统]