CoChat

Automatic Crack Classification by Exploiting Statistical Event Descriptors for Deep Learning

Open access

The DOI matches its Crossref record; 104 of the 104 references it cites were found in a registry or an index.

Abstract

In modern building infrastructures, the chance to devise adaptive and unsupervised data-driven structural health monitoring (SHM) systems is gaining in popularity. This is due to the large availability of big data from low-cost sensors with communication capabilities and advanced modeling tools such as deep learning. A promising method suitable for smart SHM is the analysis of acoustic emissions (AEs), i.e., ultrasonic waves generated by internal ruptures of the concrete when it is stressed. The advantage in respect to traditional ultrasonic measurement methods is the absence of the emitter and the suitability to implement continuous monitoring. The main purpose of this paper is to combine deep neural networks with bidirectional long short term memory and advanced statistical analysis involving instantaneous frequency and spectral kurtosis to develop an accurate classification tool for tensile, shear and mixed modes originated from AE events (cracks). We investigated effective event descriptors to capture the unique characteristics from the different types of modes. Tests on experimental results confirm that this method achieves promising classification among different crack events and can impact on the design of the future of SHM technologies. This approach is effective to classify incipient damages with 92% of accuracy, which is advantageous to plan maintenance.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.3390/app112412059

Cites

104 of 104 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 ( a ) Experimental setup. ( b ) Detail of the specimen under test with mounted sensors. ( c ) Control panel with acquired AE signals. ( d ) Experimental load-vs-time diagrams for the concrete specimens having different values of R ck ranging between 25 and 45 MPa. Inset: Schematics of the sensors layout. ( e ) A sketch of the acquisition system.
Figure 1 ( a ) Experimental setup. ( b ) Detail of the specimen under test with mounted sensors. ( c ) Control panel with acquired AE signals. ( d ) Experimental load-vs-time diagrams for the concrete specimens having different values of R ck ranging between 25 and 45 MPa. Inset: Schematics of the sensors layout. ( e ) A sketch of the acquisition system.CC-BY · From the PDF
Figure 2 Block diagram of the framework for damage classification implemented in this work.
Figure 2 Block diagram of the framework for damage classification implemented in this work.CC-BY · From the PDF
Figure 3 Examples of the time domain trace (black line), P -wave (red line) and S -wave (green line) waves as extracted with the HHT from AE waveforms related to different failure events: ( a ) tensile, ( b ) shear and ( c ) mixed-mode, respectively.
Figure 3 Examples of the time domain trace (black line), P -wave (red line) and S -wave (green line) waves as extracted with the HHT from AE waveforms related to different failure events: ( a ) tensile, ( b ) shear and ( c ) mixed-mode, respectively.CC-BY · From the PDF
Figure 4 ( a ) Time domain representation of three investigated signals representing tensile (red), shear (blue) and mixed-mode (green), respectively. ( b ) Frequency spectrum representation of the same signals.
Figure 4 ( a ) Time domain representation of three investigated signals representing tensile (red), shear (blue) and mixed-mode (green), respectively. ( b ) Frequency spectrum representation of the same signals.CC-BY · From the PDF
Figure 5 Instantaneous frequency ( a ) for tensile (red line), shear (blue line) and mixed-mode (green line) as calculated from the signals represented in Figure 4a. Plot of spectral entropy ( b ) and spectral kurtosis ( c ) as calculated on the same signals.
Figure 5 Instantaneous frequency ( a ) for tensile (red line), shear (blue line) and mixed-mode (green line) as calculated from the signals represented in Figure 4a. Plot of spectral entropy ( b ) and spectral kurtosis ( c ) as calculated on the same signals.CC-BY · From the PDF
Figure 6 Unfolded Bi-LSTM architecture with three consecutive steps ( t -1, t , t + 1).
Figure 6 Unfolded Bi-LSTM architecture with three consecutive steps ( t -1, t , t + 1).CC-BY · From the PDF
Figure 7 Proposed network for DL-based event type discrimination.
Figure 7 Proposed network for DL-based event type discrimination.CC-BY · From the PDF
Figure 8 ( a -d ) Effects of model-parameters in tuning of Bi-LSTM network with a dataset Q = 15.000 AEevents and testing of 1650 events. ( a ) Performance of training, validation and testing with different input configurations ( l 1 l 5 ) using d LSTM = 200 ( a ), d LSTM = 500 ( b ), d LSTM = 800 and ( c ), respectively. ( d ) Trend of epoch number e = 300 as a function of the different input configurations ( λ 1 λ 5 ) and for d LSTM = 200, 500 and 800, respectively.
Figure 8 ( a -d ) Effects of model-parameters in tuning of Bi-LSTM network with a dataset Q = 15.000 AEevents and testing of 1650 events. ( a ) Performance of training, validation and testing with different input configurations ( l 1 l 5 ) using d LSTM = 200 ( a ), d LSTM = 500 ( b ), d LSTM = 800 and ( c ), respectively. ( d ) Trend of epoch number e = 300 as a function of the different input configurations ( λ 1 λ 5 ) and for d LSTM = 200, 500 and 800, respectively.CC-BY · From the PDF
Figure 9 Confusion matrix of the testing performance for d LSTM = 500 and by considering the combination of IF, SE and SK as input ( l 5 ). Overall accuracy is 92%. The better intra-class accuracy is achieved on shear events (93.1%), followed by tensile event (91.9%), where the lowest is mixed-mode (90.0%). The model finds the mixed-modes most difficult to identify and primarily confuses them with actual tensile solicitations (or vice versa).
Figure 9 Confusion matrix of the testing performance for d LSTM = 500 and by considering the combination of IF, SE and SK as input ( l 5 ). Overall accuracy is 92%. The better intra-class accuracy is achieved on shear events (93.1%), followed by tensile event (91.9%), where the lowest is mixed-mode (90.0%). The model finds the mixed-modes most difficult to identify and primarily confuses them with actual tensile solicitations (or vice versa).CC-BY · From the PDF
Figure 10 ( a ) Learning error calculated by the value of loss function for training J TRN( Q ) and validation ( b ) J VAL( Q ) tasks, respectively, under different input combinations l 1 l 5 , and for different values of d LSTM.
Figure 10 ( a ) Learning error calculated by the value of loss function for training J TRN( Q ) and validation ( b ) J VAL( Q ) tasks, respectively, under different input combinations l 1 l 5 , and for different values of d LSTM.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 104 references

Research works printed but not linked: 7

  1. Barat, 2017, Acoustic signals recognition by convolutional neural network, Int. J. Appl. Eng. Res., 12, 3461
  2. Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning, The MIT Press.
  3. Lamonaca, 2010, Sensors & Transducers of Acoustic Emissions in Civil Engineering Structures by Using Time Frequency, Representation, 8, 42
  4. Munoz, D., Lara, F.B., Vargas, C., and Enriquez-Caldera, R. (2009). Position Location Techniques and Applications, Elsevier.
  5. Maddela, 2015, Spectral Kurtosis Theory: A Review through Simulations Spectral Kurtosis Theory: A Review through Simulations, Glob. J. Res. Eng. F Electr. Electron. Eng., 15, 49
  6. Ohtsu, M. (2015). Acoustic emission and related non-destructive evaluation techniques in the Ffracture mechanics of concrete, Elsevier.
  7. Bishop, C.M. (2006). Pattern recognition and machine learning. Information Science and Statistics, Springer.

Cited by

Metadata

Authors
Siracusano, G., Garescì, F., Finocchio, G., Tomasello, R., Lamonaca, F., Scuro, C., Carpentieri, M., Chiappini, M., & La Corte, A.
Affiliations
  • University of Catania (IT)
  • University of Messina (IT)
  • Polytechnic University of Bari (IT)
  • University of Calabria (IT)
  • Istituto Nazionale di Geofisica e Vulcanologia (IT)
  • Department of Electric, Electronic and Computer Engineering, University of Catania, Viale Andrea Doria 6, 95125 Catania, Italy
Corresponding author
  • Siracusano, Giulio
  • Finocchio, Giovanni
  • Chiappini, Massimo
Venue
Applied Sciences
Publisher
MDPI AG
Published
2021-12-17
Volume
11
Issue
24
Pages
12059
ISSN
2076-3417 (ISSN-L)
OpenAlex
W3217263135
Cited by
39
References
104
Citations by year
2026: 4 · 2025: 7 · 2024: 8 · 2023: 13 · 2022: 7
FWCI
3.62
Type
journal article
Language
en
Open access
Locations
  • Applied Sciences (publishedVersion, cc-by)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Archive of Doctoral Theses and Digital Collections (University of Calabria) (submittedVersion)
  • arXiv (Cornell University)
Funders
Ministry of Education, Universities and Research (J45F21000460001)
Topics
  • Physical Sciences › Engineering › Civil and Structural Engineering › Structural Health Monitoring Techniques
  • Physical Sciences › Engineering › Civil and Structural Engineering › Structural Engineering and Materials Analysis
  • Physical Sciences › Engineering › Mechanics of Materials › Ultrasonics and Acoustic Wave Propagation
Keywords
crack classification, acoustic emission, structural health monitoring, concrete damage detection, bidirectional LSTM, instantaneous frequency, spectral kurtosis, tensile cracks, shear cracks, mixed-mode cracks, maintenance planning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.