Physics-Prior and Multi-Scale Attention-Based Three-Dimensional Quality Field Reconstruction for Sparse Industrial Manufacturing Inspection
DOI:
https://doi.org/10.4108/eetsis.13131Keywords:
Intelligent manufacturing, three-dimensional defect reconstruction, industrial non-destructive testing, physics-informed neural network, multi-scale attention mechnaism, sparse inspection data, quality field reconstructionAbstract
INTRODUCTION: Three-dimensional quality field reconstruction from sparse industrial measurements is critical for intelligent manufacturing yet remains challenging due to sensor accessibility limitations, high inspection costs, and component occlusion.
OBJECTIVES: This study develops a reconstruction framework that recovers full-domain three-dimensional fields from sparse observations by integrating physics-based constraints with data-driven deep learning, and validates it on simulated three-dimensional optical scattering fields that serve as a controlled, physically rigorous proxy for industrial quality fields.
METHODS: A physics-guided loss function embedding Helmholtz equation residuals and boundary conditions is constructed, combined with a multi-scale three-dimensional convolutional network incorporating serial channel-spatial attention to capture both global structural and local high-frequency defect features. The benchmark fields are solved on a 128 × 128 × 128 grid, and 5% of voxels are retained as labeled samples.
RESULTS: Under 5% sparse sampling of the simulated optical scattering fields, the proposed framework achieved the best reconstruction performance among all compared methods. On 40 independent test samples, it reached a PSNR of 34.6 ± 1.3 dB, an SSIM of 0.932 ± 0.011, and a phase RMSE of 0.183 ± 0.025 rad, outperforming MS-CNN by 4.5 dB, 0.049, and 33.2%, respectively. Visual comparisons showed improved recovery of strong scattering regions, boundary variations, diffraction-focus positions, and high-frequency defect-related structures, with an average peak-intensity error of about 4.2% and centroid deviation of about 0.3 μm. Under noisy inputs, the method retained PSNR above 29 dB and SSIM above 0.88 at SNR = 10 dB, indicating stronger robustness than 3D U-Net and standard PINN. Ablation experiments (reported as mean ± standard deviation over the 40 test fields) further confirmed that physical constraints, multi-scale convolution, channel attention, and spatial attention each contributed to the overall performance gain, with paired statistical tests confirming significant improvements across the main comparisons.
CONCLUSION: The framework offers a robust, physically consistent route for sparse quality-field reconstruction in intelligent manufacturing. Validation currently relies on high-fidelity numerical simulation; extension to real industrial measurements is identified as the primary direction of future work.
References
[1] Ashebir D A, Hendlmeier A, Dunn M, et al. Detecting multi-scale defects in material extrusion additive manufacturing of fiber-reinforced thermoplastic composites: a review of challenges and advanced non-destructive testing techniques[J]. Polymers, 2024, 16(21): 2986.
[2] Zhang X, Dong M, Cai X, et al. Progress in machining-induced residual stress and microstructural evolution of inhomogeneous materials and composites[J]. Advanced Composites and Hybrid Materials, 2023, 6(3): 122.
[3] Usamentiaga R, Menendez F, Perez L. Towards low-cost surface inspection using artificial vision in practical industrial environments[J]. IEEE Transactions on Industry Applications, 2026.
[4] Meister S, Grundhoefer L, Stüve J, et al. Imaging sensor data modelling and evaluation based on optical composite characteristics: Investigation of data quality for inline inspection[J]. The International Journal of Advanced Manufacturing Technology, 2021, 116(11): 3965-3990.
[5] Feng F, Zhao H, E K, et al. Subaperture stitching for large-aperture wavefront detection based on modified Hartmann measurement method[J]. Optical Engineering, 2025, 64(7): 074104-074104.
[6] Teixeira F L, Sarris C, Zhang Y, et al. Finite-difference time-domain methods[J]. Nature Reviews Methods Primers, 2023, 3(1): 75.
[7] Stefanski T P. Applications of the discrete Green's function in the finite-difference time-domain method[J]. Progress In Electromagnetics Research, 2013, 139: 479-498.
[8] Penwarden M., Zhe S., Narayan A., et al. Multifidelity modeling for Physics-Informed Neural Networks (PINNs)[J]. Journal of Computational Physics, 2022, 451: 110844.
[9] Hosseini M Y, Shiri Y. Flow field reconstruction from sparse sensor measurements with physics-informed neural networks[J]. Physics of Fluids, 2024, 36(7).
[10] Colombo D, Turkoglu E, Li W, et al. Coupled physics-deep learning inversion[J]. Computers & Geosciences, 2021, 157: 104917.
[11] Oommen V., Shukla K., Goswami S., et al. Learning two-phase microstructure evolution using neural operators and autoencoder architectures[J]. npj Computational Materials, 2022, 8(1): 190.
[12] Dong Y., Chen S., Liu X.J., et al. Physics-constrained neural networks for half-space seismic wave modeling[J]. Computers and Geosciences, 2023, 181: 105477.
[13] Ronneberger O., Fischer P., Brox T. U-Net: Convolutional networks for biomedical image segmentation[C]// Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Cham: Springer, 2015: 234–241.
[14] Vinuesa R., Brunton S.L. Enhancing computational fluid dynamics with machine learning[J]. Nature Computational Science, 2022, 2(6): 358–366.
[15] Aikawa Y., Ueda N., Tanaka T. Improving the Efficiency of Training Physics-Informed Neural Networks Using Active Learning[J]. New Generation Computing, 2024, 42(4): 739–760.
[16] Raissi M., Perdikaris P., Karniadakis G.E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2019, 378: 686–707.
[17] Cai S., Mao Z., Wang Z., et al. Physics-informed neural networks (PINNs) for fluid mechanics: a review[J]. Acta Mechanica Sinica, 2021, 37(12): 1727–1738.
[18] Escapil-Inchauspé P., Ruz G.A. Hyper-parameter tuning of physics-informed neural networks: Application to Helmholtz problems[J]. Neurocomputing, 2023, 561: 126826.
[19] Xiong L., Zeng Y., Huang X., et al. MLANet: multi-level attention network with multi-scale feature fusion for crowd counting[J]. Cluster Computing, 2024, 27(5): 6591–6608.
[20] Woo S., Park J., Lee J.Y., et al. CBAM: Convolutional block attention module[C]// Proceedings of the European Conference on Computer Vision (ECCV). Cham: Springer, 2018: 3–19.
[21] Bodong Z, Maojin T, Guohua Z, et al. An Integrating Physics-Informed Neural Networks and Attention Mechanism based Intelligent Permeability Prediction Method for Shale Reservoirs[J]. SPE Journal, 2026: 1-18.
[22] Cheng W.C., Huang J., Zhang Y., et al. Mamba neural operator: Who wins? Transformers vs. state-space models for PDEs[J]. Journal of Computational Physics, 2026, 548: 114567.
[23] Mojgani R., Balajewicz M., Hassanzadeh P. Kolmogorov n–width and Lagrangian physics-informed neural networks: A causality-conforming manifold for convection-dominated PDEs[J]. Computer Methods in Applied Mechanics and Engineering, 2023, 404: 115810.
[24] Jagtap A.D., Kawaguchi K., Karniadakis G.E. Adaptive activation functions accelerate convergence in deep and physics-informed neural networks[J]. Journal of Computational Physics, 2020, 404: 109136.
[25] Liu B, Li H, Bian S, et al. A Gravity Data Denoising Method Based on Multi-Scale Attention Mechanism and Physical Constraints Using U-Net[J]. Applied Sciences, 2025, 15(14): 7956.
[26] Xu Y, Li R, Liu X, et al. Multi-Scale Dual-Attention Feature Network with Bidirectional Temporal Constraints for Tool Wear Monitoring[J]. Coatings, 2026, 16(3): 291.
[27] May M C, Oberst J, Lanza G. Managing product-inherent constraints with artificial intelligence: production control for time constraints in semiconductor manufacturing[J]. Journal of Intelligent Manufacturing, 2024, 35(8): 4259-4276.
[28] Schiassi E., Calabrò F., De Florio M., et al. Use of ELMs in Physics-Informed Machine Learning: Estimates on the Generalization Error in Learning ODE Solutions[J]. Journal of Scientific Computing, 2026, 107(1): 22.
[29] Cardoso Ferreira F.S., Patsios S., Paris P. φ-PINNs: Physics-informed neural networks on complex geometries[J]. Engineering Applications of Artificial Intelligence, 2024, 127: 107324.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Liangyu Chen
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
This is an open access article distributed under the terms of the CC BY-NC-SA 4.0, which permits copying, redistributing, remixing, transformation, and building upon the material in any medium so long as the original work is properly cited.
