摘要
为提高道岔伤损识别系统的识别率和实时性,本文提出一种基于对数功率谱标准差和顺序统计滤波的高速道岔振动信号端点检测算法。该算法以对数功率谱标准差作为轮对经过测点时有效振动信号和轮对未经过测点时噪声的鉴别特征,首先将每帧振动信号的功率谱划分为若干子带,然后把相继若干帧的子带功率谱经过顺序统计滤波获得滤波后功率谱,最后根据对数功率谱标准差对高速道岔振动信号进行端点检测。文中对道岔不同位置和不同信噪比的振动信号进行了大量的实测实验,实验结果表明了算法的有效性。
In order to improve the recognition rate and real time feature of the turnout injury recognition sys- tem, the high-speed turnout vibration signal endpoint detection algorithm based on the logarithmic power spec- trum standard deviation and order statistic filtering was proposed. It made use of the logarithmic power spec- trum standard deviation as the discrimination feature to distinguish valid vibration signals when wheels passing the measure point from noises when no wheels passing the measure point. Each frame power spectrum of the vibration signal was divided into several sub-bands at first and then an order statistic filter was applied to a se- quence of sub-band power spectra. High-speed turnout vibration signal endpoint detection was conducted ac- cording to the logarithmic power spectrum standard deviation finally. A large number of experiments were con- ducted with the vibration signals both at different turnout positions and under different SNR conditions and the results show the effectiveness of the proposed algorithm.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2013年第11期60-65,共6页
Journal of the China Railway Society
基金
铁道部科技研究开发计划(2011G009
2013X012-A-1
2013X012-A-2)
中央高校基本科研业务费(SWJ-TU11CX041
SWJTU12CX099)
关键词
高速道岔振动信号
对数功率谱标准差
顺序统计滤波
端点检测
high-speed turnout vibration signal~ logarithmic power spectrum standard deviation
order statistics filter
endpoint detection
