Structural Defect Detection Using Percussion Method And Machine Learning

In recent years,the research on the reliability and safety of civil infrastructures has attracted much attention.Meanwhile,structural health monitoring(SHM)and damage diagnostic techniques has become a fast-developing research topic in civil engineering.Most of the monitoring or detection approaches require equipped-sensors on the monitored structure to evaluate its state.However,the sensors employed in some approaches are quite expensive,and the operation of some approaches is rather tedious and time-consuming.Therefore,an easy-operated and inexpensive method is needed to solve some engineering problems.In this paper,the feasibility of the percussion method in pipeline deposition detection and rock bolts pre-tension loss detection is investigated.Some conclusions with theoretical and practical value are obtained.The major contents of this paper are concluded as follows:(1)Deposits removal in pipelines has great importance in ensuring pipeline operation.Selecting a suitable removal time or removal approach based on the composition and mass of the deposits not only reduces cost but also improves efficiency.This paper proposed a new non-destructive approach using the percussion method and voice recognition with support vector machine(SVM)to detect the sandy deposits in the steel pipeline.Particularly,as the mass of sandy deposits in the pipeline changes,the impact-induced sound signals will be different.A commonly used voice recognition feature,Mel-Frequency Cepstrum Coefficients(MFCCs),which represent the result of a cosine transform of the real logarithm of the short-term energy spectrum on a Mel-frequency scale,is adopted in this research and MFCCs are extracted from the obtained sound signals.An SVM model was employed to identify the sandy deposits with different mass values by classifying the energy summation of power spectrum density(PSD)and MFCCs.In addition,the classification accuracies of energy summation and MFCCs are compared.The experimental results demonstrated that MFCCs perform better in pipeline deposits detection and have great potential in acoustic recognition for SHM.Additionally,the proposed MFCCs-based pipeline deposits detection model can estimate the deposits in the pipeline with high accuracy.(2)As a type of reinforcing member,rock bolts are widely adopted in mining tunnels and underground construction to prevent the movement and expansion of rock strata.However,the loss of pre-tension will contribute to the rock bolts failure and further lead to economic and personnel losses.Monitoring the pre-tension loss of rock bolts and giving warnings prior to its failure is of vital importance in ensuring the safe underground operation.In this paper,a cross correlation-based pre-tension index,which represents overall changes in the frequency spectrum,was developed to evaluate the degradation level of the rock bolt pre-tension.By tapping the anchor plate using an impact hammer,it can be observed that the frequency spectrum of impact-induced sound signals indicated the pre-tension status of rock bolts.To verify the effectiveness of the proposed approach,three different anchor plates were employed in the tapping experiment.Compared with the result of MFCCs-based support vector regression,the proposed index can effectively evaluate the low pre-tension state of rock bolts.