Evaluation of bridge decks with overlays using impact echo, a deep learning approach

Dorafshan, Sattar; Azari, Hoda

doi:10.1016/j.autcon.2020.103133

Cited by 57 publications

(52 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…According to the acceleration time-history curve analysis, the amplitude at each measuring point in each working condition was different. For the stability of the convolutional neural network during the training process, a normalization layer is added to the proposed network (Dorafshan and Azari, 2020). The amplitudes of the acceleration time-histories are normalized as…”

Section: Data Collectionmentioning

confidence: 99%

“…Abdeljaber et al (2018) presented an enhanced CNN-based approach which only needs two measurement sets for structural damage detection. Dorafshan and Azari (2020) proposed onedimensional CNN which can successfully detect the subsurface defects of cement overlay bridge using impact echo data. Li et al (2018) proposed a damage-identification method for bridges based on a CNN.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Identification of Grout Sleeve Joint Defect in Prefabricated Structures Using Deep Learning

et al. 2020

View full text Add to dashboard Cite

A grout sleeve connection is a typical kind of joint in prefabricated structures. However, for construction and manufacturing reasons, defects in this kind of joint are usually inevitable. The joint quality of a prefabricated structure has a significant influence on its overall performance and can lead to structural failure. Due to the complexity of various types of materials used in grout sleeve connections, traditional non-destructive testing methods, such as Acoustic Emission (AE), Ultrasonic Testing (UT), Guided Wave Testing (GW), are facing great challenges. The recent development of deep learning technology provides a new opportunity to solve this problem. Deep learning can learn the inherent rules and abstract hierarchies of sample data, and it has a powerful ability to extract the intrinsic features of training data in complex classification tasks. This paper illustrates a deep learning framework for the identification of joint defects in prefabricated structures. In this method, defect features are extracted from the acceleration time history response of a prefabricated structure using a convolutional neural network. The proposed method is validated by vibration experiments on a half-scaled, two-floor prefabricated frame structure with column rebars spliced by different defective grout sleeves.

show abstract

Section: Data Collectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Identification of Grout Sleeve Joint Defect in Prefabricated Structures Using Deep Learning

et al. 2020

View full text Add to dashboard Cite

show abstract

“…In the majority of machine learning algorithms, user input is required to select the features considered for classification. Recently, Dorafshan et al (2020) developed a deep learning model for training convolutional neural networks to detect artificial delamination in laboratory-made reinforced concrete specimens where the features were extracted autonomously and through training [11,20]. The training datasets of the convolutional neural network (CNN) in raw IE data collected from the subsurface of structure with no pre-processing.…”

Section: Introductionmentioning

confidence: 99%

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

Jafari

Dorafshan

2021

Infrastructures

Self Cite

View full text Add to dashboard Cite

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.

show abstract

“…The impact-echo method is easy to use and does not require expensive instrumentation. Combining classification models, or deep convolutional neural networks, can be used as a powerful tool seen in this study [ 22 ]. Other applications can be found in testing concrete pavements, which is done by a laser crack measurement system [ 23 ].…”

Section: Introductionmentioning

confidence: 99%

Effect of Hammer Type on Generated Mechanical Signals in Impact-Echo Testing

Dvořák

Topolář

2021

Materials

View full text Add to dashboard Cite

The impact-echo diagnostic method is a well-known nondestructive pulse compression test method, which can be relatively easily used for the testing of concrete and reinforced concrete elements. The evaluation of the measurement with this method is based on the analysis of the signal itself in the time and frequency domains. This allows acquisition of information on the velocity of the mechanical wave, the resonant frequency of the specimen or on the presence of internal defects. The ability to interpret these measurements depends on the experience of the diagnostic technician. The advent of classification algorithms in the field of machine learning has brought an increasing number of applications where the entire interpretation phase can be considerably simplified with the help of classification models. However, this automated evaluation procedure must be provided with the information of whether the signal acquired by the test equipment has actually been measured under optimally set conditions. This paper proposes a procedure for the mutual comparison of different measuring setups with a variable tip type, hammer handle and impact force. These three variables were used for a series of measurements which were subsequently compared with each other using multi-criteria evaluation. This offers a tool for the evaluation of measured data and their filtering. As an output of the designed method, each measurement is marked by a score value, which represents how well the acquired signal fit the weight demands for each observed feature of the signal. The method allows the adjustment of selected demands for a specific application by means of set thresholds. This approach enables the understanding of characteristics of the signal in the automated pre-processing of measured data, where computing power is limited. Thus, this solution is potentially suitable for remote long-term observations with sensor arrays or for acoustic emission signals pre-processing.

show abstract

Evaluation of bridge decks with overlays using impact echo, a deep learning approach

Cited by 57 publications

References 30 publications

Identification of Grout Sleeve Joint Defect in Prefabricated Structures Using Deep Learning

Identification of Grout Sleeve Joint Defect in Prefabricated Structures Using Deep Learning

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

Effect of Hammer Type on Generated Mechanical Signals in Impact-Echo Testing

Contact Info

Product

Resources

About