Neural Network-Based Inverse Analysis for Defect Identification with Laser Ultrasonics

Oishi, Atsuya; Yamada, Kanji; Yoshimura, Shinobu; Yagawa, Genki; Nagai, Satoshi; Matsuda, Youichi

doi:10.1080/09349840109409688

Cited by 17 publications

(3 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The method was tested experimentally on one plate by measuring its thickness, Young's modulus, and Poisson's ratio. In an another previous work, dynamic responses calculated by a nite-element method were used for the network training [23].…”

Section: State Of the Art: Machine Learning For Ultrasonic Inspectionmentioning

confidence: 99%

Reduced Training Data for Robust Ultrasound Signal Interpretation by Neural Networks

Rus,

Fleury

2023

Preprint

View full text Add to dashboard Cite

We introduce an experimental setup to generate large datasets of ultrasonic signals relevant for quality inspection. A reconfigurable defect is induced by a heating laser projecting a short line on a shape memory polymer foil, which has a special property that its Young’s modulus and ultrasound attenuation level can locally be controlled by its temperature field. Ultrasound is generated by a laser pulse at one fixed position and detected by a laser vibrometer at another fixed position for 64 different defect positions and 3 different configurations of the specimen. The obtained diversified datasets are used to optimize the network architecture for the interpretation of ultrasound signals. We study the robustness of the model in cases of reduced and dissimilar training datasets. In our first study, we classify the specimen configurations with the defect position being the disturbing parameter. The model shows high performance on a dataset of signals obtained at all the defect positions, even if only trained on a completely different dataset containing signals obtained at few defect positions. In our second study, we perform precise defect localization. The model becomes robust to the changes in the specimen configuration when a reduced dataset, containing signals obtained at two different specimen configurations, is used for the training process. These conclusions show the great potential of the demonstrated machine learning algorithm for industrial quality control. High-volume products (simulated by a reconfigurable specimen in our work) can be rapidly tested on the production line using this single-point and contact-free ultrasonic method.

show abstract

Section: State Of the Art: Machine Learning For Ultrasonic Inspectionmentioning

confidence: 99%

Reduced Training Data for Robust Ultrasound Signal Interpretation by Neural Networks

Rus,

Fleury

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…In addition, Wang et al proposed an algorithm that uses the fractal dimension of crack recognition to automatically identify specific types of crack defects [7]. Although automatic discrimination systems and related algorithms based on image processing have been continuously proposed, there are still few mature industrial applications [8][9][10][11].…”

Section: Introductionmentioning

confidence: 99%

Investigation on the Formation Mechanism of Crack Indications and the Influences of Related Parameters in Magnetic Particle Inspection

Yang

Cai

et al. 2020

Applied Sciences

View full text Add to dashboard Cite

The recent rapid development of industrial cameras and machine learning has brought new vitality to the very traditional flaw detection method, namely, magnetic particle inspection (MPI). To fully develop automatic fluorescent MPI technology, two main issues need to be solved urgently—the lack of theoretical analysis on the formation of the crack indications, and quantitative characterization methods to determine the crack indications. Here, we carry out a theoretical analysis and an experimental approach to address these issues. Theoretical models of the acting force of the leakage magnetic field were established. Subsequently, the impacts of different magnetic field strengths (1000–9000 A/m) and magnetic particle concentrations (0.5–30 mL/L) on the adsorption critical distance were analyzed. The models were solved by numerical calculations in MATLAB. In addition, a single variable control experiment was conducted to study the effects of crack images. In order to determine the quality of the crack image, three characteristic parameters were investigated, such as indication gray scale, background gray scale, and contrast ratio, were provided. The theoretical magnetic particle concentration range provided a guidance value for automated fluorescent MPI. Experimental results revealed that the optimal magnetic particle concentration was 3–4 mL/L, and, under this condition, the contrast between the crack indications and the background of crack images was obvious.

show abstract

“…It was long after Horace Lamb theory about the existence of Lamb waves that for the first time Firestone and Ling 1951 succeeded in experimentally generating Lamb waves in plates [5]. Nowadays, Lamb waves can be actively generated and collected with various methods such as ultrasonic probes coupled with angle-adjustable Perspex wedges 2 International Journal of Aerospace Engineering [6]; Hertzian contact transducer [7]; laser-based ultrasonic (LBU) for Lamb wave generation and laser interferometer for their acquisition [8]; piezoelectric lead zirconate titanate (PZT) elements [9]; and interdigital transducers such as polyvinylidene fluoride (PVDF) piezoelectric polymer film [10].…”

Section: Introductionmentioning

confidence: 99%

A Novel Complementary Method for the Point-Scan Nondestructive Tests Based on Lamb Waves

Gorgin

Zheng

2014

International Journal of Aerospace Engineering

View full text Add to dashboard Cite

This study presents a novel area-scan damage identification method based on Lamb waves which can be used as a complementary method for point-scan nondestructive techniques. The proposed technique is able to identify the most probable locations of damages prior to point-scan test which lead to decreasing the time and cost of inspection. The test-piece surface was partitioned with some smaller areas and the damage probability presence of each area was evaluated.A0mode of Lamb wave was generated and collected using a mobile handmade transducer set at each area. Subsequently, a damage presence probability index (DPPI) based on the energy of captured responses was defined for each area. The area with the highest DPPI value highlights the most probable locations of damages in test-piece. Point-scan nondestructive methods can then be used once these areas are found to identify the damage in detail. The approach was validated by predicting the most probable locations of representative damages including through-thickness hole and crack in aluminum plates. The obtained experimental results demonstrated the high potential of developed method in defining the most probable locations of damages in structures.

show abstract

Neural Network-Based Inverse Analysis for Defect Identification with Laser Ultrasonics

Cited by 17 publications

References 26 publications

Reduced Training Data for Robust Ultrasound Signal Interpretation by Neural Networks

Reduced Training Data for Robust Ultrasound Signal Interpretation by Neural Networks

Investigation on the Formation Mechanism of Crack Indications and the Influences of Related Parameters in Magnetic Particle Inspection

A Novel Complementary Method for the Point-Scan Nondestructive Tests Based on Lamb Waves

Contact Info

Product

Resources

About