Acoustic Vibration Of Concrete Pavement Void Detection Signal Processing Method

Entered into a new century, the cement concrete pavements has obtained a substantialdevelopment, the total of the cement pavement has reached250,000km, accounting for18%of the total mileage of the various grades closing the end of the2004. Cement concretepavement is a pavement structure of stiffness, strong load diffusion capacity and good stability,and it has been more widely used because that it is perfect in technically and has been anadvantage in economic. But when it has been used for a period of time, there came all kindsof diseases, and void under the board was one of the common disease. In this paper, first haveunderstudied comprehensively and analyzed the void detection technology at home andabroad and acoustic signal processing method, and then have full used of the goodperformance of the MFCC and HHT in the road void feature extraction,combined with theoutstanding performance of the BP neural network in automatic recognition, last have putforward a signal processing method of the detection of void status of the concrete pavementbased on sound vibration method for the specific characteristics of the acoustic signalscollected by acoustic vibration.In the paper, described the acoustic vibration detection theory and the signal processingprocess of the acoustic vibration detection. According to the formation mechanism of itssound, put forward signal analysis methods for this kind of the signal based on itscharacteristics. The approach mainly included the time-domain analysis and thetime-frequency analysis, in the time-domain analysis, detected endpoint of signal based onshort-term average amplitude; in the time-frequency analysis, put forward short-time Fouriertransform and Hilbert-Huang transform and studied its theoretical. And put forward a schemefor signal feature extracting by making use of MFCC coefficients and HHT, which have madetargeted improvements based on a large number of experiments.In this paper, in order to realize the accurate discrimination void of the concrete pavement,established a network model which consisted of MFCC and Hilbert marginal spectrum byin-depth studying and researching of BP neural network, and then discrimination experimentwas processed for void signal.