摘要
混凝土路面在使用过程中会出现各种病害,板下脱空就是常见的病害之一。因此,对混凝土路面板下脱空的有效识别就显得十分必要和紧迫。文中在了解和分析国内外脱空识别和声信号处理方法的基础上,针对声振法采集到的声信号的具体特征,利用梅尔倒谱系数(MFCC)和BP神经网络在混凝土路面板下脱空程度方面的应用,对路面板下脱空进行了研究。结果表明:严重脱空的识别正确率为94%,一般性脱空的识别正确率为92%,非脱空的识别正确率为90%。本文研究方法能够达到识别路面板下脱空的目的。
There are all sorts of diseases on the use of the concrete pavement,such as the empty beneath the concrete pavement slab. Hence,the effective identification for the empty beneath the concrete pavement slab is very necessary and urgent. This paper studies the empty beneath the concrete pavement slab,on the basis of a comprehensive understanding and analysis of identifying the empty beneath the concrete pavement slab and the acoustical signal processing method at home and abroad,for the specific characteristics of the signal acquisition based on the acoustic method,in making good use of the applications in terms of the degree of the empty beneath the concrete pavement slab for Mel-scale Frequency Cepstral Coefficient( MFCC) and BP neural network. The results demonstrate that the identification accuracy ratio of severe empty,general empty and nonempty is 94%,92% and90% respectively. The method of this paper can reach the goal of identifying the empty beneath the concrete pavement slab.
作者
朱宇
王楠
Zhu Yu;Wang Nan(Department of Computer Technology and Applications,Qinghai University,Xining 810016,China;TBEA Xi’an Electric Technology Co.,Ltd,Xi’an 710000,China)
出处
《青海大学学报(自然科学版)》
2018年第3期27-33,39,共8页
Journal of Qinghai University(Natural Science)
基金
青海大学中青年科研基金项目(2014-QGY-27)
Google科研培育项目
关键词
梅尔倒谱系数
声振法
板下脱空
神经网络
Mel-scale Frequency Cepstral Coefficient
acoustic vibration method
empty beneath concrete pavement slab
neural network