Automated Defect Recognition for Welds Using Simulation Assisted TFM Imaging with Artificial Intelligence

Gantala, Thulsiram; Balasubramaniam, Krishnan

doi:10.1007/s10921-021-00761-1

Cited by 31 publications

(16 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Scaling the technique to work with ultrasonic arrays over a selected RoI developed the total focusing method (TFM). 12 This has been a widely used 13,14 and optimized 15 technique for ultrasonically evaluating materials for defects. Configuring the TFM algorithm with different travel modes like halfskip 16,17 and full skip have also been performed in detection of notches in metals with high-frequency ultrasound.…”

mentioning

confidence: 99%

An ultrasonic wave‐based framework for imaging internal cracks in concrete

Kuchipudi

Ghosh

2022

Structural Contr & Hlth

View full text Add to dashboard Cite

Nondestructive detection and sizing of internal cracks initiated in reinforced concrete have been critical problems. Geometric characterization of internal cracks and delaminations contributes to predicting their propagation path and failure pattern followed by structural prognostics. There have been limited studies on quantitative detection and characterization of vertical cracks in concrete using wave-based NDE techniques. This study focuses on solving the problem of imaging deep internal crack planes in the concrete medium by leveraging the ultrasonic shear horizontal (SH) waves from a transducer array. Postprocessing the array full matrix capture (FMC) data with both total focusing method (TFM) and plane wave imaging (PWI) reveals their efficiency in mapping planar defects inclined between 0 and 60 . However, their performance on vertical/near-vertical defects is found to be inferior with imaging limited to the tip of the defect planes. Further, half-skip modes of wave dispersion have been adopted in addition to the directly scattered pulses for imaging the vertical and near-vertical cracks. We propose an imaging framework based on the outcomes of our investigation on the best suitable methodology to detect and map planar defects like cracks inclined in the range of 0-90 . Statistical quantification of defect inclinations from the reconstructed images is compared to the ground-truth orientations, and they are found to be 94%-99% accurate. Besides accuracy, the computational efficiency of the proposed techniques makes them desirable and reliable for quick on-site inspections on built infrastructure.full matrix capture (FMC), total focusing method (TFM), half-skip total focusing method (HSTFM), plane wave imaging (PWI), shear horizontal (SH) waves, ultrasonic array, vertical cracks | INTRODUCTIONInternal cracks in reinforced concrete are a result of various factors including thermal, mechanical, and environmental deterioration. Cracking affects the performance of structural concrete at different stages in its lifecycle. Diagnosis of such defects is followed by retrofitting at early stages to mitigate the failure of structures. The non-homogeneity and highly attenuating nature of concrete medium poses a challenge for the nondestructive community to evaluate structures for thin underlying cracks.

show abstract

mentioning

confidence: 99%

An ultrasonic wave‐based framework for imaging internal cracks in concrete

Kuchipudi

Ghosh

2022

Structural Contr & Hlth

View full text Add to dashboard Cite

show abstract

“…Our responses are shown in italics: About ultrasonic ray-tracing, the authors state: "The solution can be approximated by classical methods, e.g. by finite element method or boundary element method 21,28 . "…”

Section: Ensure Full Reproducibility? Partlymentioning

confidence: 99%

“…The authors should address the following questions/issues: About ultrasonic ray-tracing, the authors state: "The solution can be approximated by classical methods, e.g. by finite element method or boundary element method 21,28 ." The authors seem to not be aware of recent solutions that have been proposed to solve the ultrasonic ray-tracing problem using iterative root-finding methods.…”

Section: Acknowledgementsmentioning

confidence: 99%

Training neural networks on domain randomized simulations for ultrasonic inspection

Schlachter¹,

Felsner²,

Zambal³

2022

Open Res Europe

View full text Add to dashboard Cite

To overcome the data scarcity problem of machine learning for nondestructive testing, data augmentation is a commonly used strategy. We propose a method to enable training of neural networks exclusively on simulated data. Simulations not only provide a scalable way to generate and access training data, but also make it possible to cover edge cases which rarely appear in the real world. However, simulating data acquired from complex nondestructive testing methods is still a challenging task. Due to necessary simplifications and a limited accuracy of parameter identification, statistical models trained solely on simulated data often generalize poorly to the real world. Some effort has been made in the field to adapt pre-trained classifiers with a small set of real world data. A different approach for bridging the reality gap is domain randomization which was recently very successfully applied in different fields of autonomous robotics. In this study, we apply this approach for ultrasonic testing of carbon-fiber-reinforced plastics. Phased array captures of virtual specimens are simulated by approximating sound propagation via ray tracing. In addition to a variation of the geometric model of the specimen and its defects, we vary simulation parameters. Results indicate that this approach allows a generalization to the real world without applying any domain adaptation. Further, the trained network distinguishes correctly between ghost artifacts and defects. Although this study is tailored towards evaluation of ultrasound phased array captures, the presented approach generalizes to other nondestructive testing methods.

show abstract

“…Although the method shows AUC of 88.4% for defect segmentation, lack of defect classification is discernible. Gantala and Balasubramaniam [45] presented an automatic defect recognition model trained on total focusing method (TFM) imaging dataset and finite element simulated dataset with addition of noise and further expansion of dataset utilizing deep convolutional gaN (DCGAN). eir two-class defect detection model was evaluated with yolov4 [46] and reached 85 average precision (AP) on the noisy dataset.…”

Section: Introductionmentioning

confidence: 99%

Real‐Time Explainable Multiclass Object Detection for Quality Assessment in 2‐Dimensional Radiography Images

Naddaf-Sh

Zargarzadeh

et al. 2022

Complexity

View full text Add to dashboard Cite

Quality inspection and defect detection play a critical role in infrastructure safety and integrity specially when it comes to aging infrastructure mostly owned by governments around the world. One of the prevalent inspections performed in the industry is nondestructive testing (NDT) using radiography imaging. Growing demand, shortage of experts, diversity of required skills, and specific regional standards with a time-limited requirement of inspection results make automated inspection an urgent need. Therefore, utilizing artificial intelligence- (AI-) based tools as an assistive technology has become a trend for industrial applications, which automates repeated tasks and provides increased confidence before and during the inspection operation. Most of the works in quality assessment are focused on the classification of few categories of defects and mostly performed on public or noncomprehensive research datasets. In this work, a scalable, efficient, and real-time deep learning family of models for detection and classification of 10 various categories of weld characteristics on a real-world industrial dataset is presented. The models are evaluated and compared against each other, various critical hyperparameters and components are optimized, and local explainability of models is discussed. Additionally, AutoAugment for object detection and various techniques are utilized and investigated. The best performance for object detection and classification for 10 class models is reached by mean average precision of 72.4% and top-1 accuracy of 90.2%, respectively. Also, the fastest object detection model is able to evaluate a full 15360 × 1024 pixels weld image in 0.39 seconds. Finally, the proposed models are deployable on edge-devices to perform as assistant to NDT experts or auditing professionals.

show abstract

Automated Defect Recognition for Welds Using Simulation Assisted TFM Imaging with Artificial Intelligence

Cited by 31 publications

References 20 publications

An ultrasonic wave‐based framework for imaging internal cracks in concrete

An ultrasonic wave‐based framework for imaging internal cracks in concrete

Training neural networks on domain randomized simulations for ultrasonic inspection

Real‐Time Explainable Multiclass Object Detection for Quality Assessment in 2‐Dimensional Radiography Images

Contact Info

Product

Resources

About