Pith. sign in

REVIEW 3 major objections 6 minor 60 references

A Machine Learning Approach to Generate Residual Stress Distributions using Sparse Characterization Data in Friction-Stir Processed Parts

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A U-Net trained on process simulations can reconstruct the full-field residual stress distribution in a friction-stir processed part from only nine measured stress locations.

desk verdict A promising demonstration that U-Net can map sparse stress measurements to full-field residual stress maps, but the experimental error metrics likely include the nine input points, so the central 10x-reduction claim is not yet rigorously quantified. read the letter →

arxiv 2506.08205 v1 pith:IWKQVW7F submitted 2025-06-09 cs.LG cs.CE

Shadab Anwar Shaikh , Kranthi Balusu , Ayoub Soulami This is my paper
classification cs.LGcs.CE
keywords residualstressfrictionstirprocessingU-NetsparsedatareconstructionholedrillingESPIthermalpseudo-mechanicalmodelfull-fieldpredictionA380aluminumalloy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that a machine-learning model can infer the full two-dimensional residual stress field left by friction stir processing from only nine measured stress values. If true, it removes the main practical barrier to residual-stress-aware design: full-field characterization currently requires dense hole drilling at roughly 120 locations, each taking about 25 minutes. The authors train a U-Net called the Residual Stress Generator only on Thermal Pseudo-Mechanical process simulations, then test it on an actual two-pass friction-stir processed A380.0 sheet. The model reproduces simulated fields to within about 1 MPa on average and reaches $R^2=0.6$ with a mean absolute error of 10.59 MPa on the experimental sample, which the authors argue is enough to justify a more than 10x reduction in measurement effort.

What carries the argument

The carrying object is the Residual Stress Generator: a U-Net with three encoder and three decoder blocks, skip connections, and a final 1×1 convolution, trained on 400 input-output pairs from Thermal Pseudo-Mechanical (TPM) simulations. The input is a 128×128 grid divided into nine sub-regions, each filled with the measured stress value at a predefined sampling location; the output is the 128×128 full residual stress map. The TPM training data come from a moving heat-flux model, $q(r,T)=\zeta\omega r\,\sigma_{\mathrm{yield}}(T)/\sqrt{3}$, with tool speed, rotation speed, and calibration factor $\zeta$ varied over wide ranges, and simulations outside roughly 400–500 °C excluded. The U-Net's encoder-decoder structure with skip connections is what lets a few point values expand into a coherent full field.

What would settle it

Run the trained RSG on a fresh friction-stir sample whose full residual stress field is measured densely, feed only the nine predefined point values, and compare predicted and measured stresses everywhere; the claim fails if overall error is not in the reported range (MAE around 10 MPa, $R^2$ near 0.6) or if systematic overprediction at plunge/exit and 5 mm outside the process zone persists.

Watch

Extended reading notes

Core claim

The central claim is that a single U-Net, trained only on simulated friction-stir residual stress fields, can act as a residual stress generator: given the thickness-averaged longitudinal stress at nine predefined locations, it outputs a 128×128 full-field map that matches the true stress field closely enough to replace dense experimental characterization. On simulated test data the generator achieves average RMSE 0.92 MPa and SSI 0.981; on an unseen simulation set it averages RMSE 4.43 MPa; and on an experimental two-pass A380.0 sample it reaches $R^2=0.6$ with $\mathrm{MAE}=10.59$ MPa using only nine input points instead of the 120 measured locations. The paper interprets this as evidence that the model has learned the latent structure of friction-stir residual stress distributions, and that the approach can cut characterization effort by more than an order of magnitude.

Load-bearing premise

The load-bearing premise is that the single-pass Thermal Pseudo-Mechanical simulations, with their fitted calibration factor and steady-state heat-flux assumption, produce residual stress fields faithful enough to the real two-pass friction-stir process that a network trained only on those simulations can generalize to experimental data.

Editorial extensions

If this is right

  • Full-field residual stress maps for friction-stir processed sheets can be obtained from nine characterization points rather than roughly 120, a more than 10x reduction in experimental effort.
  • The RSG reproduces the qualitative features known for FSP residual stress fields: tensile stresses inside and near the processed zone, highest values at the zone edges, and compressive stresses away from the processed zone.
  • On unseen simulation parameter sets the model degrades only moderately (average RMSE 4.43 MPa, SSI 0.85), indicating that the learned mapping generalizes beyond the training parameter grid.
  • Experimental prediction errors concentrate in three identifiable regions—plunge/exit, the second-pass start, and 5 mm outside the processed zone—so improving the TPM model in those regions is a direct route to higher accuracy.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The fixed nine-point input grid is an assumption, not a result: a natural next test is whether adaptive or random sensor placements, or fewer than nine points, preserve the reported accuracy.
  • Because the network learns the latent structure of a particular simulation family, the same architecture should transfer to other manufacturing processes (welding, casting, additive) provided a comparable set of process simulations exists; the paper does not test this.
  • The reported experimental error is computed after nearest-neighbor interpolation from the 128×128 grid to measurement points, so evaluating predictions directly at element centroids could change the error statistics.
  • Training on multi-pass simulations or adding plunge/exit-aware heat-source terms would likely remove the largest systematic overpredictions and improve on the experimental $R^2=0.6$.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. This paper proposes a U-Net-based Residual Stress Generator (RSG) that maps sparse residual stress measurements at nine predefined locations to a full 128x128 field of thickness-averaged longitudinal residual stress in friction-stir processed aluminum sheets. The model is trained on 400 Thermal Pseudo-Mechanical (TPM) finite-element simulations covering a range of process parameters, tested on 40 held-out simulations, evaluated for generalization on 149 additional simulations outside the training parameter range, and finally applied to a two-pass experimental A380.0 sample on which 120 ESPI hole-drilling measurements were taken. The authors report low errors on simulated fields and MAE=10.59 MPa, RMSE=13.44 MPa, and R2=0.6 on the experimental sample, and conclude that the RSG can replace 120 characterization locations with 9, a more-than-10x reduction in experimental effort.

Significance. If the central claim survives a masked re-evaluation, the approach is practically valuable: it offers a route to full-field residual stress maps from sparse hole-drilling measurements, and the experimental generalization from single-pass simulations to a two-pass sample is a demanding and meaningful test. The paper has clear strengths: a clean supervised learning setup, a separate generalization set, an independent experimental benchmark, and a candid discussion of the sources of the largest errors (plunge/exit modeling, single-pass versus two-pass differences, clamping conditions, and suspected measurement anomalies). However, the quantitative support for the headline 'more than 10x reduction' claim is currently undermined by the evaluation protocol, so the significance is conditional on the requested re-analysis with the input locations excluded from the error metrics.

major comments (3)
  1. [§3.4, §2.2.3, Table 2] The experimental validation statistics include the nine input locations. The input to the RSG is a 128x128 array in which nine sub-regions are populated with the measured stresses at the predefined sampling locations (Section 2.2.3), and Section 3.4 then reports MAE=10.59 MPa, RMSE=13.44 MPa, and R2=0.6 over the 120 characterized locations without stating that these nine input locations were excluded. Because the model is given the answer at those locations, including them in the error computation conflates memorized input values with genuine predictions. The paper's central claim that 9 measurements replace 120 requires the error on the 111 truly unmeasured locations, and that number is not reported. The same masking issue affects the simulated test metrics in Table 2, where the reference output contains the input sub-regions. Please recompute all reported metrics after excluding the nine input sub-regions, for both the simulated and experimental datasets, and report the errors separately for the input and non-input locations.
  2. [§3.4, §4] The experimental evaluation does not report measurement uncertainty for the ESPI hole-drilling technique, and it provides no baseline comparison. Without error bars on the measured stresses or repeated measurements, it is difficult to judge whether MAE=10.59 MPa is acceptable predictive accuracy or within experimental noise. In addition, the paper motivates the ML approach by contrasting it with eigenstrain inversion and other sparse-reconstruction methods, but it does not compare the RSG against a simple interpolation baseline (for example, inverse-distance or kriging using the same nine points) or against an eigenstrain inversion on this dataset. Adding measurement uncertainty and at least one baseline would substantially strengthen the claim that the approach significantly reduces experimental effort.
  3. [§2.1, §3.4, §4] The assumption that a one-pass simulation captures all essential features needed to train the RSG is load-bearing, because the experimental validation is performed on a two-pass sample. Section 2.1 asserts this without supporting evidence, and Section 4 later attributes part of the experimental error to the single-pass versus two-pass difference. Please provide a quantitative assessment of this mismatch, for example by simulating the two-pass configuration or comparing one-pass and two-pass stress fields, or explicitly temper the claim that the RSG predicts the experimental field with the stated accuracy.
minor comments (6)
  1. [§2.1] The text says 'Two 8-node cuboid temperature-displacement (DC2D4) mesh elements' but DC2D4 is a 4-node two-dimensional element; please correct the element type and description.
  2. [§3.4] In the sentence describing stress variation in the transverse direction, 'up to 5 m beyond it' should read 'up to 5 mm beyond it'.
  3. [Table 2] The header 'RSME' should be 'RMSE'.
  4. [Abstract and keywords] The phrase 'Friction Weld Processing (FSW)' appears to be a typo for 'Friction Stir Welding (FSW)'; please correct it.
  5. [§3.4] The error percentages are stated twice in slightly different forms ('over 55% of locations' with 'approximately 70% of these' below 6 MPa, later '38.5% of the locations' below 6 MPa); please present the breakdown once, with the base percentages explicit.
  6. [Data Availability] The data availability statement says data will be made available upon request; providing the trained model and preprocessing code as well would improve reproducibility.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the RSG is a learned map from sparse measurements to full fields; the reported experimental error may include input locations, but no prediction is forced by construction.

full rationale

Step-by-step, the claimed chain is: (1) TPM simulations generate residual-stress fields; (2) each field is subsampled at nine points to form the input, with the full field as target; (3) a U-Net is trained to minimize MSE between predicted and reference fields; (4) the trained network is applied to nine measured ESPI values from an independent two-pass A380.0 sample; (5) the full predicted field is compared to 120 measured locations. No equation in the paper defines the output as equal to the input, and no parameter is fitted to the experimental sample. The input is a coarse, nine-block version of the field, and the output is the synthesized high-resolution field; that is the intended inpainting task, not a tautology. The simulation training data do rely on the TPM model and the zeta calibration factor from prior co-author work [1][39], and the one-pass-versus-two-pass mismatch is a fidelity limitation; however, these are not circular because the experimental test is external and the reported degradation (Table 3 versus Table 2, and the Section 4 outlier analysis) shows the model is not memorizing the experiment. The one substantive caveat is that Section 3.4 reports MAE=10.59 MPa, RMSE=13.44 MPa, R2=0.6 over 'the characterization locations' without stating that the nine input locations were excluded; because those nine values are placed directly into the input array, their error contribution is not a pure prediction. This is a test-protocol/statistical limitation that weakens the quantitative 10x-reduction claim, but it is not a circular derivation: the other 111 locations and the full spatial structure are still synthesized by the network. No self-citation chain or uniqueness argument forces the result. Accordingly, no significant circularity is present.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the fidelity of the simulation data (TPM model with a fitted calibration factor and single-pass assumption) and on the chosen sparse sampling locations. These are not independently evidenced in this paper; a reader must accept the simulation as a proxy for real residual stress fields.

free parameters (3)
  • Calibration factor zeta (TPM heat flux model) = 0.1 to 0.5 in increments of 0.05 in training simulations; calibrated in prior work
    Equation (1) uses zeta to scale the moving heat flux; the simulations that produce all training/test targets depend on this fitted factor, so the latent stress structure learned by the RSG inherits its value range.
  • Film coefficient at backing plate = 1000 W/m^2.C
    Chosen in Section 2.1 to represent heat dissipation through the backing plate; not calibrated against experiments in this paper.
  • Sparse sampling locations (9 coordinates) = (0.074,0.010), (0.074,0.080), (0.074,0.170), (0.082,0.010), (0.082,0.080), (0.082,0.170), (0.092,0.010)…
    Input positions in Section 2.2.3 are hand-picked; the reconstruction quality depends on this choice, and no analysis of alternative sampling layouts is given.
assumptions (5)
  • domain assumption The Thermal Pseudo-Mechanical (TPM) heat-source model with Equation (1) adequately represents friction stir processing for residual stress simulation.
    Section 2.1 relies on this model to generate all training data; prior validation is cited as references 1 and 39, but the model cannot capture plunge/exit transients, first-pass hardening, or some clamping effects, as acknowledged in Section 4.
  • domain assumption A single-pass simulation captures the essential features of the two-pass experimental sample's residual stress distribution.
    Stated in Section 2.1 and used in Section 3.4; the experimental inaccuracy near the ends of the processing line is attributed to this mismatch.
  • standard math Grid interpolation of 2839 point-stresses onto a 128x128 grid faithfully preserves the fields.
    Section 2.2.3 uses SciPy grid interpolation; the resolution loss and interpolation artifacts are not analyzed, but it is a standard numerical step.
  • domain assumption Symmetry across the processing line allows a half-plate model.
    Section 2.1 models half the plate; the experimental sample is two-pass in opposite directions, so full-field symmetry is imperfect.
  • domain assumption Yield stress values in the Appendix for A380 are accurate inputs.
    Temperature-dependent yield stress feeds Equation (1); errors here propagate into all simulated stress fields.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Machine Learning Approach to Generate Residual Stress Distributions using Sparse Characterization Data in Friction-Stir Processed Parts." pith.science (2026). https://pith.science/paper/IWKQVW7F

@misc{pith2026250608205,
  author       = {Pith},
  title        = {Pith review of: A Machine Learning Approach to Generate Residual Stress Distributions using Sparse Characterization Data in Friction-Stir Processed Parts},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IWKQVW7F}},
  note         = {Machine review of arXiv:2506.08205}
}
read the original abstract

Residual stresses, which remain within a component after processing, can deteriorate performance. Accurately determining their full-field distributions is essential for optimizing the structural integrity and longevity. However, the experimental effort required for full-field characterization is impractical. Given these challenges, this work proposes a machine learning (ML) based Residual Stress Generator (RSG) to infer full-field stresses from limited measurements. An extensive dataset was initially constructed by performing numerous process simulations with a diverse parameter set. A ML model based on U-Net architecture was then trained to learn the underlying structure through systematic hyperparameter tuning. Then, the model's ability to generate simulated stresses was evaluated, and it was ultimately tested on actual characterization data to validate its effectiveness. The model's prediction of simulated stresses shows that it achieved excellent predictive accuracy and exhibited a significant degree of generalization, indicating that it successfully learnt the latent structure of residual stress distribution. The RSG's performance in predicting experimentally characterized data highlights the feasibility of the proposed approach in providing a comprehensive understanding of residual stress distributions from limited measurements, thereby significantly reducing experimental efforts.

Figures

Figures reproduced from arXiv: 2506.08205 by the authors.

Figure 1
Figure 1. Finite element model for FSP of an A380 sample [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Schematic of RSG (U-Net) architecture used in this study [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Top-view of the two-pass friction stir processed A380 sample. The vertices of the grid indicate the [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: The hole-drilling electronic speckle pattern interferometry (ESPI) setup showing all the modules [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The half-model showing a simulated thickness-averaged longitudinal stress field used for training [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Predictions obtained using the trained RSG on the test dataset [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 6
Figure 6. Figure 6: Predictions obtained using the trained RSG model on the test dataset [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Sample predictions obtained using the trained RSG on the unseen dataset [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 7
Figure 7. Figure 7: Sample predictions obtained using the trained RSG on the unseen dataset [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Results form the trained RSG using the characterization data: (a) Displays the average residual stresses at [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 8
Figure 8. Figure 8: Results form the trained RSG using the characterization data:(b) Illustrates the predicted versus measured [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

60 extracted references · 54 canonical work pages

  1. [60]

    Martínez

    Hayley Weir, Keiran Thompson, Amelia Woodward, Benjamin Choi, Augustin Braun, and Todd J. Martínez. ChemPix: automated recognition of hand-drawn hydrocarbon structures using deep learning.Chemical Science, 12(31):10622–10633, 2021. 20

    work page 2021

  2. [1]

    Kranthi Balusu, Kyoo Sil Choi, Hrishikesh Das, Avik Samanta, Piyush Upadhyay, Saumyadeep Jana, and Ayoub Soulami. On the Utility of the Thermal-Pseudo Mechanical Model’s Residual Stress Prediction Capability for the Development of Friction Stir Processing.The International Journal of Advanced Manufacturing Technology, (0123456789), mar 2023

    work page 2023

  3. [2]

    Baumann, Nilesh Kumar, and Rajiv S

    John A. Baumann, Nilesh Kumar, and Rajiv S. Mishra. Residual stresses in friction stir welding residual stresses in friction stir welding. 2014

    work page 2014

  4. [3]

    Residual stress evaluation in api 5l x65 girth welded pipes joined by friction welding and gas tungsten arc welding.Journal of Materials Research and Technology, 8(1):988–95, 2019

    Carlos Alexandre Pereira De Moraes et al. Residual stress evaluation in api 5l x65 girth welded pipes joined by friction welding and gas tungsten arc welding.Journal of Materials Research and Technology, 8(1):988–95, 2019

    work page 2019

  5. [4]

    R. John, K. V . Jata, and K. Sadananda. Residual stress effects on near-threshold fatigue crack growth in friction stir welds in aerospace alloys.International Journal of Fatigue, 25(9-11):939–48, 2003. 17 A Machine Learning Approach to Generate Residual Stress Distributions using Sparse Characterization Data in Friction-Stir Processed Parts

    work page 2003

  6. [5]

    Vinay Kumar Yadav, Vidit Gaur, and I. V . Singh. Combined effect of residual and mean stresses on fatigue behavior of welded aluminum 2024 alloy.International Journal of Fatigue, 155:2021, September 2022

    work page 2024

  7. [6]

    Chantikul, B

    P. Chantikul, B. R. Lawn, and D. B. Marshall. Micromechanics of flaw growth in static fatigue: Influence of residual contact stresses.Journal of the American Ceramic Society, 64(6):322–25, 1981

    work page 1981

  8. [7]

    W. Woo, Z. Feng, X. L. Wang, and S. A. David. Neutron diffraction measurements of residual stresses in friction stir welding: A review.Science and Technology of Welding and Joining, 16(1):23–32, 2011

    work page 2011

Show all 60 references
  1. [8]

    Angular distortion and through-thickness residual stress distribution in the friction-stir processed 6061-t6 aluminum alloy.Materials Science and Engineering A, 437(1):64–69, 2006

    Wanchuck Woo et al. Angular distortion and through-thickness residual stress distribution in the friction-stir processed 6061-t6 aluminum alloy.Materials Science and Engineering A, 437(1):64–69, 2006

    2006

  2. [9]

    Kranthi Balusu, Avik Samanta, Shivakant Shukla, Hrishikesh Das, Piyush Upadhyay, and Saumyadeep Jana. A comparative analysis of residual stresses from friction stir processing of aluminum cast 380 and wrought 7075 alloy sheets : experimental characterization and modeling.The I...

    2024

  3. [10]

    McDowell

    Yan Wang and David L. McDowell. Uncertainty quantification in materials modeling. InUncertainty Quantifica- tion in Multiscale Materials Modeling, pages 1–40. Elsevier, 2020

    2020

  4. [11]

    Material flow visualization during friction stir welding using high-speed x-ray imaging

    Hemant Agiwal et al. Material flow visualization during friction stir welding using high-speed x-ray imaging. Manufacturing Letters, 34:62–66, 2022

    2022

  5. [12]

    Yuh J. Chao, X. Qi, and W. Tang. Heat transfer in friction stir welding - experimental and numerical studies. Journal of Manufacturing Science and Engineering, Transactions of the ASME, 125(1):138–45, 2003

    2003

  6. [13]

    H. B. Schmidt and J. H. Hattel. Thermal modelling of friction stir welding.Scripta Materialia, 58(5):332–37, 2008

    2008

  7. [14]

    Song and R

    M. Song and R. Kovacevic. Thermal modeling of friction stir welding in a moving coordinate system and its validation.International Journal of Machine Tools and Manufacture, 43(6):605–15, 2003

    2003

  8. [15]

    Coupling Smoothed Particle Hydrodynamics With Finite Element Method to Simulate Residual Stresses From Friction Stir Processing

    Kranthi Balusu, Lei Li, Kyoo Sil Choi, and Ayoub Soulami. Coupling Smoothed Particle Hydrodynamics With Finite Element Method to Simulate Residual Stresses From Friction Stir Processing. InVolume 9: Mechanics of Solids, Structures, and Fluids; Micro- and Nano-Systems Engineeri...

    2022

  9. [16]

    Escobar, Hrishikesh Das, Shivakant Shukla, Benjamin J

    Lei Li, Julian D. Escobar, Hrishikesh Das, Shivakant Shukla, Benjamin J. Schuessler, Nicole R. Overman, Glenn J. Grant, Suveen N. Mathaudhu, Jorge F. Dos Santos, Cynthia A. Powell, and Ayoub Soulami. A dislocation density-based meshfree computational framework for solid phase ...

    2024

  10. [17]

    P. J. Withers and H. K.D.H. Bhadeshia. Residual stress part 1 - Measurement techniques.Materials Science and Technology, 17(4):355–365, 2001

    2001

  11. [18]

    Paul S. Prevey. X-ray diffraction residual stress techniques. InMaterials Characterization. ASM International, 01 1986

    1986

  12. [19]

    The measurement of residual stresses by the incremental hole drilling technique

    P V Grant, J D Lord, and P S* Whitehead. The measurement of residual stresses by the incremental hole drilling technique. Measurement good practice guide, National Physical Laboratory (NPL), Teddington, 2006

    2006

  13. [20]

    G. S. Schajer and M. Steinzig. Full-field calculation of hole drilling residual stresses from electronic speckle pattern interferometry data.Experimental Mechanics, 45(6):526–32, 2005

    2005

  14. [21]

    Micromechanics of defects in solids, 1987

    Tanaka Mura. Micromechanics of defects in solids, 1987

    1987

  15. [22]

    Jazeel Rahman Chukkan et al. An iterative technique for the reconstruction of residual stress fields in a butt-welded plate from experimental measurement, and comparison with welding process simulation.International Journal of Mechanical Sciences, 160:421–28, July 2019

    2019

  16. [23]

    A. M. Korsunsky. Residual elastic strain due to laser shock peening: Modelling by eigenstrain distribution.The Journal of Strain Analysis for Engineering Design, 41(3):195–204, 2006

    2006

  17. [24]

    Korsunsky

    Tea Sung Jun and Alexander M. Korsunsky. Evaluation of residual stresses and strains using the eigenstrain reconstruction method.International Journal of Solids and Structures, 47(13):1678–86, 2010

    2010

  18. [25]

    Finite element analysis and machine learning guided design of carbon fiber organosheet-based battery enclosures for crashworthiness.Applied Composite Materials, 0123456789, 2024

    Shadab Anwar Shaikh et al. Finite element analysis and machine learning guided design of carbon fiber organosheet-based battery enclosures for crashworthiness.Applied Composite Materials, 0123456789, 2024

    2024

  19. [26]

    Probabilistic surrogate model for accelerating the design of electric vehicle battery enclosures for crash performance.arXiv preprint arXiv:2408.03450, 2024

    Shadab Anwar Shaikh, Harish Cherukuri, Kranthi Balusu, Ram Devanathan, and Ayoub Soulami. Probabilistic surrogate model for accelerating the design of electric vehicle battery enclosures for crash performance.arXiv preprint arXiv:2408.03450, 2024. 18 A Machine Learning Approac...

    2024 arXiv

  20. [27]

    Prediction of composite microstructure stress- strain curves using convolutional neural networks.Materials & Design, 189:108509, 2020

    Charles Yang, Youngsoo Kim, Seunghwa Ryu, and Grace X Gu. Prediction of composite microstructure stress- strain curves using convolutional neural networks.Materials & Design, 189:108509, 2020

    2020

  21. [28]

    Stress field prediction in fiber-reinforced composite materials using a deep learning approach.Composites Part B: Engineering, 238:109879, 2022

    Anindya Bhaduri, Ashwini Gupta, and Lori Graham-Brady. Stress field prediction in fiber-reinforced composite materials using a deep learning approach.Composites Part B: Engineering, 238:109879, 2022

    2022

  22. [29]

    Data-driven methods for stress field predictions in random heterogeneous materials.Engineering Applications of Artificial Intelligence, 123:106267, 2023

    Enjamamul Hoq, Osama Aljarrah, Jun Li, Jing Bi, Alfa Heryudono, and Wenzhen Huang. Data-driven methods for stress field predictions in random heterogeneous materials.Engineering Applications of Artificial Intelligence, 123:106267, 2023

    2023

  23. [30]

    Improved deep learning method for accurate flow field reconstruction from sparse data.Ocean Engineering, 280:114902, 2023

    Jinsong Zhang, Jian Liu, and Zhenwei Huang. Improved deep learning method for accurate flow field reconstruction from sparse data.Ocean Engineering, 280:114902, 2023

    2023

  24. [31]

    Robust flow reconstruction from limited measurements via sparse representation.Physical Review Fluids, 4(10):103907, 2019

    Jared L Callaham, Kazuki Maeda, and Steven L Brunton. Robust flow reconstruction from limited measurements via sparse representation.Physical Review Fluids, 4(10):103907, 2019

    2019

  25. [32]

    A practical approach to flow field reconstruction with sparse or incomplete data through physics informed neural network.Acta Mechanica Sinica, 39(3):322302, 2023

    Shengfeng Xu, Zhenxu Sun, Renfang Huang, Dilong Guo, Guowei Yang, and Shengjun Ju. A practical approach to flow field reconstruction with sparse or incomplete data through physics informed neural network.Acta Mechanica Sinica, 39(3):322302, 2023

    2023

  26. [33]

    Fully convolutional networks for semantic segmentation

    Evan Shelhamer, Jonathan Long, and Trevor Darrell. Fully convolutional networks for semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(4):640–51, 2017

    2017

  27. [34]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. InMedical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part I...

    2015

  28. [35]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High-resolution image synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022

    2022

  29. [36]

    Schmidt and J

    H. Schmidt and J. Hattel. A local model for the thermomechanical conditions in friction stir welding.Modelling and Simulation in Materials Science and Engineering, 13(1):77–93, 2005

    2005

  30. [37]

    H. B. Schmidt and J. H. Hattel. Thermal modelling of friction stir welding.Scripta Materialia, 58(5):332–337, 2008

    2008

  31. [38]

    Abaqus 6.14 CAE User Guide

    Dassault Systèmes Simulia. Abaqus 6.14 CAE User Guide. page 1146, 2014

    2014

  32. [39]

    On the utility of the thermal-pseudo mechanical model ’ s residual stress prediction capability for the development of friction stir processing on the utility of the thermal

    Kranthi Balusu et al. On the utility of the thermal-pseudo mechanical model ’ s residual stress prediction capability for the development of friction stir processing on the utility of the thermal. 2022

    2022

  33. [40]

    P. L. Threadgilll, A. J. Leonard, H. R. Shercliff, and P. J. Withers. Friction stir welding of aluminium alloys. International Materials Reviews, 54(2):49–93, 2009

    2009

  34. [41]

    Denseunet: densely connected unet for electron microscopy image segmentation.IET Image Processing, 14(12):2682–2689, 2020

    Yue Cao, Shigang Liu, Yali Peng, and Jun Li. Denseunet: densely connected unet for electron microscopy image segmentation.IET Image Processing, 14(12):2682–2689, 2020

    2020

  35. [42]

    Visual geometry group-unet: deep learning ultrasonic image reconstruction for curved parts.The Journal of the Acoustical Society of America, 149(5):2997–3009, 2021

    Yujian Mei, Haoran Jin, Bei Yu, Eryong Wu, and Keji Yang. Visual geometry group-unet: deep learning ultrasonic image reconstruction for curved parts.The Journal of the Acoustical Society of America, 149(5):2997–3009, 2021

    2021

  36. [43]

    Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J

    Pauli Virtanen, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, E...

    2020

  37. [44]

    Seth T Strayer, William J Frieden Templeton, Florian X Dugast, Sneha P Narra, and Albert C To. Accelerating high-fidelity thermal process simulation of laser powder bed fusion via the computational fluid dynamics imposed finite element method (cifem).Additive Manufacturing Let...

    2022

  38. [45]

    A deep learning approach for the velocity field prediction in a scramjet isolator.Physics of Fluids, 33(2), 2021

    Chen Kong, Juntao Chang, Yunfei Li, and Ziao Wang. A deep learning approach for the velocity field prediction in a scramjet isolator.Physics of Fluids, 33(2), 2021

    2021

  39. [46]

    Topology optimization using super- resolution image reconstruction methods.Advances in Engineering Software, 177:103413, 2023

    Seunghye Lee, Qui X Lieu, Thuc P V o, Joowon Kang, and Jaehong Lee. Topology optimization using super- resolution image reconstruction methods.Advances in Engineering Software, 177:103413, 2023

    2023

  40. [47]

    Image quality metrics: Psnr vs

    Alain Hore and Djemel Ziou. Image quality metrics: Psnr vs. ssim. In2010 20th international conference on pattern recognition, pages 2366–2369. IEEE, 2010. 19 A Machine Learning Approach to Generate Residual Stress Distributions using Sparse Characterization Data in Friction-S...

    2010

  41. [48]

    Multiscale structural similarity for image quality assessment

    Zhou Wang, Eero P Simoncelli, and Alan C Bovik. Multiscale structural similarity for image quality assessment. InThe Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, volume 2, pages 1398–1402. Ieee, 2003

    2003

  42. [49]

    Martín Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S. Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Ian Goodfellow, Andrew Harp, Geoffrey Irving, Michael Isard, Yangqing Jia, Rafal Jozefowicz, Lukasz Kaiser, Manjunat...

    2015

  43. [50]

    Ma- chine learning–accelerated computational fluid dynamics.Proceedings of the National Academy of Sciences, 118(21):e2101784118, 2021

    Dmitrii Kochkov, Jamie A Smith, Ayya Alieva, Qing Wang, Michael P Brenner, and Stephan Hoyer. Ma- chine learning–accelerated computational fluid dynamics.Proceedings of the National Academy of Sciences, 118(21):e2101784118, 2021

    2021

  44. [51]

    Grant, and Saumyadeep Jana

    Avik Samanta, Hrishikesh Das, Glenn J. Grant, and Saumyadeep Jana. Effect of tool design and pass strategy on defect elimination and uniform, enhanced tensile properties of friction stir processed high-pressure die-cast A380 alloy.Materials Science and Engineering: A, 861(Octo...

    2022

  45. [52]

    Residual Stress Measurement by ESPI Hole-Drilling.Procedia CIRP, 45:203–206, 2016

    Theo Rickert. Residual Stress Measurement by ESPI Hole-Drilling.Procedia CIRP, 45:203–206, 2016

    2016

  46. [53]

    Steinzig and T

    M. Steinzig and T. Takahashi. Residual stress measurement using the hole drilling method and laser speckle interferometry part III: Analysis Technique.Experimental Techniques, 27(6):59–63, 2003

    2003

  47. [54]

    G. S. Schajer. Application of Finite Element Calculations to Residual Stress Measurements.Journal of Engineering Materials and Technology, 103(2):157–163, apr 1981

    1981

  48. [55]

    G. S. Schajer. Full-field Calculation of Hole Drilling Residual Stresses from Electronic Speckle Pattern Interfer- ometry Data.Experimental Mechanics, 45(6):526–532, dec 2005

    2005

  49. [56]

    Steinzig and E

    M. Steinzig and E. Ponslet. RESIDUAL STRESS MEASUREMENT USING THE HOLE DRILLING METHOD AND LASER SPECKLE INTERFEROMETRY: PART 1.Experimental Techniques, 27(3):43–46, may 2003

    2003

  50. [57]

    Schönberger, Juan Nunez-Iglesias, François Boulogne, Joshua D

    Stéfan van der Walt, Johannes L. Schönberger, Juan Nunez-Iglesias, François Boulogne, Joshua D. Warner, Neil Yager, Emmanuelle Gouillart, Tony Yu, and the scikit-image contributors. scikit-image: image processing in Python.PeerJ, 2:e453, 6 2014

    2014

  51. [58]

    Kranthi Balusu, Hrishikesh Das, Shivakant Shukla, Ayoub Soulami, and Piyush Upadhyay. Investigating strength increasing modifications to a friction stir welded AA7075 T-peel joint’s exit location: a modeling led study.The International Journal of Advanced Manufacturing Technol...

    2024

  52. [59]

    Reinforcing the Exit Hole from Friction Stir Welding and Processing.SSRN Electronic Journal, 26(August):101611, 2022

    Tianhao Wang, Tingkun Liu, Timothy Roosendaal, and Piyush Upadhyay. Reinforcing the Exit Hole from Friction Stir Welding and Processing.SSRN Electronic Journal, 26(August):101611, 2022

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.