Identification And Classification Of Concrete In Time-Frequency Analysis Based On Microphone Array Audio Signal

Cement concrete pavement with strong carrying capacity,high temperature,long life and other irreplaceable advantages,is widely used in road construction and occupies a certain proportion in China's road transport system.However,it is often threatened by the bottom of the plate,the bottom of the cement concrete pavement is one of the most common hazards,seriously affecting the safety of transportation,so the cement concrete pavement bottom blank detection is necessary.The traditional nondestructive examination method adopts single microphone to receive sound signal,whose anti-reference ability is poor and the spatial resolution is low.In order to overcome the shortcoming of classical nondestructive examination method,microphone array is proposed to receive audio signals in this paper.It is easy to produce cross terms to analysis the time-frequency using Wigner-Ville transform.In order to deal with the problem,Born-Jordan transform is proposed to suppress the cross terms of sound signal in this paper.This paper is use of microphone array to collect audio signals,use of broadband signal theory of the collected signal for sound source positioning and beam formation;The Born-Jordan time-frequency analysis of the audio signal is carried out and the effective features are extracted,and the extraction features are reduced dimension by KPCA;SVM classification is carried out to evaluate the post-dimensioning characteristics and to evaluate the voiding condition of cement concrete pavement.Based on the measured data,this paper uses MATLAB to simulate the proposed method.The experimental results show that the accuracy of the proposed method can reach 87.5%,which is 4.69 percentage points higher than that of the traditional acoustic vibration non-destructive testing method.The method proposed in this paper is feasible,basically able to accurately determine the road of the void.