Bridge Inspection and Defect Recognition with Using Impact Echo Data, Probability, and Naive Bayes Classifiers

Abstract

Interpretation of IE data have been carried out by analyzing IE signals in frequency domain to determine the maximum frequency. However, the current peak frequency method can be inaccurate. The purpose of this research is to introduce features in IE signals that can be used for effective classification and interpretation for bridge deck evaluation through statistical analysis and Naive Bayes classifiers. The dataset contained IE data collected from eight slabs created at Advanced Sensing Technology FAST NDE laboratory (FHWA). A set of statistical features in time domain, normalized peak values, and length of preprocessed signals were used to classify the IE data, statistically. Then, Naive Bayes classifiers was employed to recognize defect area. Finally, the result of statistical classification was compared with frequency approach. The result shows that 19 and 21% of the IE signals collected from the defect area have multiple peaks, respectively. However, 85% of the IE signals collected from the sound set had only one peak. A probability classifier was used to find the relationship between the result of the frequency method and statistical analysis. The result shows that 10% of the IE signals were usable for estimating the thickness in the sound group.

SDNET2021: Annotated NDE Dataset for Subsurface Structural Defects Detection in Concrete Bridge Decks

“…Additionally, the peak frequency method is prone to error itself. It has resulted in false delamination detection in the past (Jafari et al, 2021). The comparative analysis in this section shows the importance of having an annotated IE dataset for more accurate delamination detection.…”

“…IE signals were preprocessed using the traditional peak frequency and Fast Fourier Transform (FFT) approach as a preliminary processing technique for IE dataset validation in comparison to generated ground truth [38]. The defect maps from the analysis of the raw IE data showed some similarity with the ground truth (Figure 14b).…”

Comparison between Supervised and Unsupervised Learning for Autonomous Delamination Detection Using Impact Echo

“…Figure 2 depicts the filtering process on a typical IE signal. A frequency analysis was used to interpret the IE signals; however, the absolute and loca peak signal values in the time domain could also be employed to classify the data [28] The mathematical algorithm used to detect distinguishable IE signal peaks is illustrated in Figure 2a. The first and last points, less than 0.1 of the absolute maximum points, would lead to removing the left and right signal sections.…”

“…On average, the summation from removing normalized points is approximately zero, which does not impact data classification. The authors of [28] provided more details on the filtering ap proach. A threshold of 0.1 was defined as the index of trivial point removal.…”

“…The performance of the developed models was also evaluated using the IE data of a bridge deck available on InfoBridge. Combining signal processing with other health monitoring devices, such as UAVs, to monitor structure health has been the subject of much current research [26][27][28][29]. Fascista et al reviewed UAV applications and assisted structural health monitoring to monitor bridge health conditions [26] and address using drones for infrastructure inspection.…”