Additive manufacturing (AM) enables integrated one-piece fabrication of parts, high material utilization efficiency, and unparalleled design freedom. However, problems such as low production efficiency, difficulties in ensuring quality stability and defect control limit the large-scale industrial application of AM. In-situ active modulation for AM enables dynamic regulation of parts during the fabrication process, thereby enhancing the quality of the final fabricated parts without introducing extra processing steps. In-situ active regulation enables direct intervention during defect nucleation, providing better effectiveness than post-printing repairs while avoiding performance degradation risks associated with post-processing. Based on the difference of core factors directly affected during regulation, in-situ active regulation is categorized into the following. (1) Process and path parameter optimization, where regulation directly impacts manufacturing-related procedural rules. It is the simplest method of control and the preferred approach, with widespread attention focused on its effects on microstructure and mechanical properties. (2) Laser beam shaping, where regulation directly influences the energy carrier morphology. To address issues such as edge over-melting and localized energy deficiency caused by non-uniform energy distribution, laser beam shaping should be employed. (3) Additional physical field modulation achieved by superimposing supplementary physical fields. When optimal process and path parameters still fail to obtain the desired microstructure and mechanical properties, additional physical field control may be considered. Meanwhile, this work summarized the effects of different additional physical fields on the mechanical properties of various metallic base materials. The future trends of in-situ modulation in additive manufacturing are also discussed.

The wire arc additive manufacturing (WAAM) of aluminium alloy leads to residual tensile stresses and porosity, which adversely affect mechanical properties. Laser shock peening (LSP) is widely used to enhance the properties of WAAM components; this study aims to elucidate the mechanisms by which LSP without coating (LSPwC) influences the microstructure and properties of WAAM aluminium alloy. The results demonstrate that the LSPwC process employs the surface metal of the aluminium alloy as a sacrificial layer, ionizing it to form plasma. Under the influence of a water confinement layer, dynamic shock waves are generated within the material, effectively introducing a residual compressive stress of-92.5 MPa at the surface of the WAAM aluminium alloy. Furthermore, LSPwC results in high-density dislocation tangles, fine slip bands, dislocation cells, and dislocation walls. The dislocation density in the impacted region significantly increased, from 1.68 & times; 1013 (m-2) to 3.41 & times; 1013 (m-2), while the porosity content markedly decreased, by 29.4%. The LSPwC aluminium alloy exhibited increases in Vickers hardness, tensile strength, and high-cycle fatigue life of 43.8%, 7.8% and 157.9%, respectively, compared to the as-deposited aluminium alloy. Therefore, the LSPwC process effectively enhances both the static strength and dynamic fatigue performance of WAAM aluminium alloys by introducing residual compressive stress and promoting dislocation proliferation. These findings highlight the effectiveness of LSPwC, providing valuable insights for improving the performance of complex geometries.

Laser Repair (LR) via Directed Energy Deposition (DED) is a key additive manufacturing technology for high-value applications. However, LR introduces microstructural heterogeneity and interfaces that complicate fatigue assessment, especially in hydrogen-rich environments which exacerbate fatigue damage. This study employs the crack tip localization method and Crack Opening Rate (COR) parameter via BSL 3D DIC to investigate hydrogen-accelerated fatigue crack propagation in laser-repaired (LR) GH4169 superalloy. The fatigue lives and crack closure levels of pure substrate (PS), pure deposited (PD), and LR specimens were compared. Hydrogen charging significantly accelerated crack growth, reducing fatigue life by 23

The application of laser directed energy deposition (LDED) for the in-situ repair of key GH4169 superalloy components confronts two technical barriers: fatigue performance degradation in the repaired region and extreme service conditions. To overcome these limitations, systematic characterization of high-temperature fatigue behavior is essential for assessing repair quality and informing process optimization. Among the controllable LDED parameters, laser power (LP) directly affects fabrication quality and plays a significant role in influencing the fatigue crack growth (FCG) behavior. In this study, the high-temperature bi-prism-based single-lens three-dimensional digital image correlation (BSL 3D DIC) system was employed to investigate the FCG behavior of compact tension specimen types under varying LPs (500/650/800 W) and temperatures (23 °C/650 °C), including pure substrate, pure deposited, and laser repaired (LR) specimens. The fatigue test for the LR specimens was set up to allow the crack to propagate along the repair interface. The results show that an LP of 650 W restored the fatigue life of the LR specimen to a level comparable to that of the pure deposited material, whereas powers outside this range significantly reduced the performance. A novel parameter, crack opening ratio, is introduced to characterize the crack closure effect. The evolution characteristics of the crack closure effect throughout the entire high-temperature fatigue process for repaired specimens under different LPs are presented and comparatively analyzed, clarifying the influence of LPs on FCG behavior. The Paris’ law incorporating the crack closure effect is thus modified, offering a robust framework for process optimization and high-temperature fatigue life prediction in LR complex structures.

In photomechanics, speckles and gratings serve as essential carriers or sensors for deformation measurement in methods including digital image correlation (DIC) and Moiré. Their fabrication quality directly influences the measurement accuracy, especially in extreme conditions, such as high-temperature and micro/nano-scale. At high temperatures, deformation carriers are prone to oxidation, degradation and ablation; concurrently, fabricating high quality and parametrically defined deformation carriers at miniature scales presents a significant challenge. This review describes recent progress and limitations in the fabrication of deformation carriers for extreme conditions, emphasizing high-temperature and micro/nano-scale. A classification and discussion of classical and state-of-the-art techniques is presented, evaluating them based on parametric fabrication capability, and their destructive or non-destructive characteristics. In addition, fabrication techniques for other extreme conditions including impact, underwater, and cryogenic temperature, are synthesized and analyzed. This review aims to provide a theoretical foundation and methodological reference for fabricating and optimizing deformation carriers, enabling reliable deformation measurements in extreme conditions.

Laser repair technology has emerged as a critical additive manufacturing solution for restoring geometries and enhancing performance of damaged high-value components like aero-engine blades. Predicting their fatigue life is essential for evaluating service reliability, particularly for components operating in extreme environment, such as high-temperature. However, conventional fracture mechanics models struggle to quantify the nonlinear relationship between processing parameters and fatigue life. Meanwhile, traditional data-driven machine learning approaches are hindered by the lengthy timelines and high costs associated with high-temperature fatigue experiments, resulting in limited datasets and a lack of physical interpretability. To address these issues, this paper proposes a Low-Data-Consumption Physics-Informed Meta-Learning model. By integrating the statistical distribution laws of fatigue life to construct a physics-informed regularization term, expanding the dataset via the Monte Carlo method, and employing a meta-learning framework to extract knowledge shared across tasks, this approach enhances generalization capability under conditions of limited data. Using only 49 experimental data points grouped into four distinct sets based on process parameters for cross-validation, this model achieved reliable prediction of the high-temperature fatigue life probability distribution for laser-repaired GH4169 specimens. Validation results demonstrate that, on average, only two predicted life value fell outside the 2-fold scatter bound, indicating strong predictive stability and accuracy. Furthermore, compared with the conventional Gaussian Process Regression model, the standard deviation predicted by this model is more consistent with the true distribution of the data, allowing it to provide more reliable uncertainty estimation.

Full-field displacement measurement is essential for understanding material deformation, validating numerical simulations, and ensuring structural safety. Conventional optical methods usually require artificial speckle patterns or gratings, which are difficult to apply and unreliable under microscale observation, high-temperature conditions, high-speed rotation, or enclosed environments. To address these limitations, this study proposes TeXNet, an unsupervised deep learning network for full-field displacement prediction on natural texture surfaces. A multi-scale attention-based feature fusion mechanism is designed to handle weak contrast, repetitive structures, and multi-scale texture variations, while photometric consistency, structural similarity, and smoothness regularization are jointly employed for end-to-end unsupervised training. Validation on synthetic datasets, as well as macroscopic carbon fiber reinforced polymer (CFRP) translation and microscopic nickel-based single-crystal superalloy (NBSC) tensile experiments, demonstrates that TeXNet can stably produce continuous and consistent displacement fields without manual annotation, showing strong robustness and engineering applicability.

High-precision three-dimensional (3D) dynamic pose measurement and tracking are of crucial value in applications such as condition monitoring of rotating machinery, robotic navigation, and assembly and docking of aircrafts. To address the limitations of traditional visual measurement methods in dynamic environments, including noise susceptibility, feature instability, and insufficient real-time performance, an optimized digital encoded markers and their recognition and localization method combining deep learning and binocular vision are proposed in this study to achieve high-precision 3D pose measurement and tracking. In particular, the recognition network, constructed using the RefineDet framework and enhanced by an automated sample generation strategy based on real-world scenarios, significantly improves recognition performance. The integration of multiple image processing algorithms such as Sobel operator, Non-Maximum Suppression (NMS), and double threshold algorithm further enables precise marker recognition and center localization, providing a reliable foundation for 3D reconstruction and pose measurement. Experiments with color-coded markers under complex backgrounds and lighting conditions, as well as tests with markers arranged across multiple spatial planes, indicate that the method exhibits excellent generalizability and stability. The condition monitoring experiment about the rotating blades verified that the optimized coded markers have high encoding capacity, excellent anti-interference ability and fast recognition performance, and meanwhile, the proposed method accurately measured key parameters, including blade pose, flapping angle, and common taper, demonstrating outstanding precision and robustness. This approach provides reliable technological support for pose measurement, tracking, and fault diagnosis in complex dynamic systems.

Thermal-field-assisted additive manufacturing (AM) has been demonstrated as an effective approach for reducing residual stress and improving the overall performance of metallic components. However, to achieve high-quality fabrication, the processing technology window is the critical challenge that must be addressed, and its practical implementation still relies heavily on trial-and-error experiments or computationally expensive finite element (FE) simulations. To facilitate the optimization of thermal-field-assisted additive manufacturing processes, this study proposes a Gaussian process regression (GPR)-based framework for the rapid prediction and optimization of residual stress fields in laser directed energy deposition (LDED). The influence of different scanning strategies, including line scanning, serpentine scanning, and out-in scanning, on the temperature and stress fields was systematically investigated. Based on a comparative analysis, the line-scanning strategy was selected for subsequent modeling. Laser power and substrate preheating temperature were identified as the key process parameters, and 32 representative parameter combinations were generated using Latin hypercube sampling. Corresponding calibrated high-fidelity FE simulations were performed to construct the training dataset. A thermal-field-assisted GPR model was then developed, enabling efficient and accurate prediction of both temperature fields and residual stress distributions under limited sample conditions. Furthermore, the proposed framework was applied to single-objective optimization of residual stress and multi-objective optimization based on Pareto fronts. Under different weight schemes, the optimal combination of process parameters was obtained, achieving a coordinated balance between residual stress reduction and thermal-history-related material considerations. Finally, the thermo-mechanical evolution mechanisms were systematically analyzed by leveraging the GPR-predicted temperature and stress fields, thereby revealing the intrinsic roles of key process parameters in residual stress generation and regulation during AM. The findings obtained in this work provide theoretical insight and practical guidance for intelligent process regulation in LDED.

Based on digital image correlation technology, the fatigue crack growth behavior and crack closure effect in laser repaired GH4169 superalloy under varying process parameters and spatial locations are investigated by using the modified Paris law and crack opening ratio, and their underlying mechanisms are elucidated through electron backscatter diffraction analysis. The results reveal that the residual stresses in laser repaired GH4169 superalloy significantly influence crack closure effect during fatigue crack growth. Moreover, the functionally graded laser repair processes improve fatigue life while reducing metallurgical defects. The grain size of GH4169 superalloy predominantly governs the threshold stress intensity factor range, whereas its effect on FCG rates is less significant. The metallurgical defects (such as lack of fusion, pores, and unmelted particles), secondary cracking, and grain boundary-induced crack deflection significantly increase both crack growth rate and data scatter. The findings in this work provide theoretical guidance and technical support for tailoring laser repair strategies to enhance fatigue resistance in engineering components.

Laser Directed Energy Deposition (LDED) has played a significant role in the additive remanufacturing applications for repairing key industrial components. However, the extreme environment and complex industrial process involved in LDED present considerable challenges to the quality and performance of repaired products, which calls for advanced in-situ online monitoring methods, especially the full-field optical metrology to address the issue. In response to the relevant needs, a novel polarized coherent gradient sensor (P-CGS) is proposed in this paper, which can achieve spatially phase-shifting efficiently and realize in-situ online deformation measurement in industrial scenarios with high accuracy. Polarization state of the object wave is considered with Jones matrix to derive the new measurement principle and governing equations. Then, an integrated P-CGS insitu online deformation measurement system is developed with polarization gratings and a pixelated polarization camera. The accuracy of the proposed method and system is verified by standard experiments with spherical reflector and four-point bending beam. In application, the developed system was used to characterize the fullfield deformation evolution of repaired components online during the LDED repair process. The time-series evolution of the out-of-plane displacement gradient, curvature, and topography in the entire repair process was obtained and analyzed in detail. In summary, the presented P-CGS method greatly improves the efficiency as it achieves high-accuracy real-time deformation measurement without additional phase-shifting device, which provides a new way for in-situ online mechanical analysis during the complex additive manufacturing process.

In-situ characterization techniques fundamentally break through the static limitation of traditional ex-situ methods, which only capture the initial and final state data of materials, by real-time tracking of the dynamic evolution of microstructures under multi-field coupling environments. This article systematically reviews five major in-situ technique families, including in-situ X-ray diffraction (XRD)/synchrotron radiation and computed tomography (CT), digital image correlation (DIC), scanning electron microscopy (SEM), electron backscatter diffraction (EBSD), and transmission electron microscopy (TEM). It elucidates the measurement principles, complementary advantages, and applicable boundaries of these techniques from the lattice/atomic scale to the engineering component level. The review highlights typical applications of these techniques in addressing key scientific problems such as load partitioning in multiphase alloys, three-dimensional evolution of damage, phase transformation kinetics, and fatigue fracture. Furthermore, it deeply analyzes four core challenges currently faced: the inherent trade-off between spatial and temporal resolution, difficulties in reproducing extreme service environments, the pressure of processing massive multimodal data, and the non-uniqueness dilemma in the inversion of microscopic model parameters. This work aims to provide a critical fundamental theoretical foundation for establishing the dynamic “processing–structure–property” relationship of materials.

During laser-directed energy deposition (L-DED), the three-dimensional morphology and surface flow behavior of the melt pool is considered as critical physical fields which fundamentally determine the quality of fabricated components. Nevertheless, conventional in-situ monitoring systems have limitations in achieving synchronized observation of both morphological and flow characteristics. In this study, we develop an in-situ optical measurement system based on four-mirror module, enabling binocular vision imaging of the melt pool using a single high-speed camera. By incorporating deep learning-assisted feature extraction and matching algorithms, three-dimensional reconstruction of the melt pool morphology is accomplished. Simultaneously, digital image correlation techniques are employed to quantify surface flow fields. The dynamics evolution of melt pool morphology and surface flow are characterized under different process parameter combinations. On this basis, surrogate models correlating process parameters with melt pool characteristics are established. Sobol sensitivity analysis further reveals the influence of process parameters on the characteristics of melt pool. The obtained results demonstrate that the melt pool height is predominantly regulated by scanning speed and powder feed rate, whereas the melt pool width exhibits stronger dependence on laser power and scanning speed. The melt pool surface flow exhibits unsteady state, with turbulent kinetic energy concentrating predominantly in the central region. Marangoni convection is identified as the dominant mechanism governing melt transport along the scanning direction. This work provides a robust experimental framework for investigating multi-physics coupling phenomena in L-DED melt pools, offering technical support for process optimization and closed-loop control strategies of L-DED process.

Abstract The traditional coherent gradient sensing (CGS) method requires the assumption of small-deformation bending, and high measurement error may occur with increasing deformation. To tackle this issue, a novel geometrically nonlinear CGS (GN-CGS) method is proposed in this study to improve measurement accuracy under large deformation. Dimensionless deformation coefficients for quantifying the systematic error of traditional CGS are defined and analyzed using parametric simulation. The results of this show that an additional 1% systematic error will be introduced into the measurement results of traditional CGS when the out-of-plane displacement gradient deformation is higher than 4 × 10 −2 ; thus its influence on deformation from the measurement error should be taken into account. For spherical and parabolic surfaces in engineering applications, the corresponding simplified measurement equations are deduced and the measurement accuracy exhibited through fringe simulation. In order to verify the feasibility of the proposed GN-CGS method, an experiment is conducted using a spherical mirror with a curvature of 0.5 ± 0.0005 m −1 . Our results indicate that the relative error of topography, out-of-plane displacement gradient and curvature using the proposed GN-CGS method are within 1% of the nominal value overall, with accuracy improved by 36.68%, 5.52% and 3.25%, respectively, compared with traditional CGS. The proposed GN-CGS method could lay a solid foundation for further applications of CGS to large-deformation scenarios.

In metal additive manufacturing, accurate monitoring and measurement of multi-physics fields during molten pool evolution are crucial for defect mechanism analysis and quality control. However, current monitoring techniques are limited to two-dimensional (2D) and single-physics observations, hindering the simultaneous acquisition of multi-physics information, particularly in three dimensions. To address these challenges, this study proposes an online in situ monitoring system for capturing multi-physics fields of the molten pool. Comprehensive molten pool information is obtained through dual-waveband, dual-view synergistic imaging. By combining two-color thermometry, intelligent identification and tracking of unmolten powders, the SETC (Speckle, Epipolar geometry, Temperature consistency, and Cluster) multi-constraint matching method, and data fusion techniques, the system enables, for the first time, single-camera synchronized measurement of highresolution 2D temperature and flow fields, as well as local three-dimensional (3D) multi-physics fields of the molten pool. Validation using cylinder experiments demonstrates high accuracy, with radius error below 0.74 %, a mean point cloud error of 0.025 mm and a mean displacement error of 0.004 mm. In Ti-6Al-4V powder printing, two-view 2D temperature and flow fields of the molten pool were captured by the system. Furthermore, local 3D information and pattern analysis of the molten pool were realized based on the identification, tracking and reconstruction of the unmolten powders. With its compact design, affordability, tunable optical pathways, high accuracy, and comprehensive data acquisition, the system offers an efficient approach for online in situ monitoring, high-precision measurement, and analysis of coupling mechanisms in the multi-physics fields of the molten pool during additive manufacturing.

Quantitative characterization of the mechanical response during dynamic shear rupture is crucial for understanding the physical mechanisms of earthquake generation. Recent advances in laboratory-scale fault simulations have enhanced this understanding by integrating optical techniques such as digital image correlation, photoelasticity, and real-time contact observation. While each method has specific strengths and limitations, combining them for comprehensive rupture analysis remains a challenge due to measurement discrepancies. In this study, the stress and displacement fields were simultaneously determined through the combined application of photoelasticity and the sample moire method. Transparent photoelastic models embedded with gratings were fabricated using three-dimensional (3D) printing technology, satisfying the requirements of both techniques. To capture the evolution of stress and displacement fields around the fault throughout the entire rupture process, time-series photoelastic fringe patterns with superimposed gratings were reconstructed along lines located above and below the fault surface. From these time-series images, the principal stress difference was directly obtained via photoelastic fringes, while the displacement and velocity fields were extracted using the sampling moire method. The dynamic evolution of both the fringe patterns and displacement fields offers new insight into the analysis of dynamic shear rupture processes. Key rupture characteristics, including non-uniform slip velocity, stress distributions behind the rupture front, the identification of both initial and subsequent rupture tips, and rupture propagation velocities, were quantitatively determined and examined. The results demonstrate that the proposed combined method offers a novel and effective approach for characterizing the mechanical behavior associated with dynamic shear rupture.

The grating phase analysis method, as an approach based on the deformation field of gratings, has been widely used for measuring both out-of-plane and in-plane displacements on specimen surfaces. However, when being applied to the dynamic deformation measurement, the computing process relates to large amounts of image data, which leads to high computation cost and limited efficiency. With the recent progress in artificial intelligence technique, deep learning has emerged as promising alternatives to overcome these limitations. In this study, a high-precision two-dimensional in-plane displacement prediction method for moiré gratings based on deep learning is proposed. A dual-branch convolutional neural network, named by GratingNet, is constructed to effectively fuse spatial-domain and frequency-domain features. The network integrates deformable convolutions, attention mechanisms, and multi-scale feature extraction modules to enable effective prediction of in-plane displacement fields using deep analysis of pre- and post-deformation images. Furthermore, an uncertainty-weighted multi-task loss function is employed to achieve adaptive balancing between spatial and frequency domain learning tasks, enhancing the robustness of the model. The proposed method is validated with the star-shaped displacement fields and randomly generated displacement field data with varying control point intervals, respectively. The results demonstrate that GratingNet has better performance than the existing methods. Compared with the classical U-Net network, GratingNet achieves better displacement measurement accuracy, with the mean absolute error (MAE) reduced by 55.03% and 54.02% in the x-direction (u) and y-direction (v), respectively, while the root mean square error (RMSE) is reduced by 52.61% and 46.74%, respectively. In comparison with the traditional Sampling Moiré method, GratingNet exhibits higher measurement efficiency, achieving a 90.72% improvement in computational efficiency. These results indicate that the proposed method is well-suited for in-plane displacement field characterization, offering high computational efficiency and prediction accuracy.

Research on the performance of Carbon Fiber Reinforced Polymer (CFRP) under low-temperature conditions is significant. This study developed a novel microscopic cooling deformation measurement system based on the sampling moire method, the system integrates the advantages of the sampling moire method for measurements and holistic design optimization for low-temperature environments. It features a simple process, high precision, and high stability, and is suitable for various materials. The system primarily consists of a cooling chamber, antifreeze circulation device, laser confocal microscope, and temperature controller, among other components. The temperature regulation range of the cooling platform spans from -40 degrees C to 150 degrees C. This study used CFRP [0 degrees/90 degrees/0 degrees] as the specimen, and the constructed system was employed to measure microscopic deformation during cooling and heating processes. The results revealed that deformation along the CFRP's longitudinal direction is minimal, whereas, in the thickness direction, the 0 degrees layer primarily dictates the deformation behavior. The 0 degrees layer consistently constrains the deformation of the 90 degrees layer, and shear strain exhibits significant variations at the interface. This research presents the first application of the sampling moire method in lowtemperature deformation measurement. It elucidates the effects of varying temperatures on the interfacial mechanical properties of CFRP while demonstrating the suitability of the developed system for low-temperature microscopic deformation measurements of specimens.

Laser repair (LR) is an important branch of additive manufacturing and one of the core technologies in strategic emerging industries like equipment remanufacturing. Fatigue failure analysis of LR parts in service is the basis for an in-depth application of LR technology in aerospace, etc. The advanced digital image correlation (DIC) method can provide strong support due to its advantages such as full-field, non-contact, and in-situ deformation measurement. However, the extreme service environments and the non-uniformity of core LR structures bring challenges. This paper investigates the fatigue crack growth behavior of the LR nickel-based superalloy GH4169 at 650 degrees C using the high-temperature bi-prism-based single-lens (BSL) 3D DIC system, which primarily comprises a double telecentric lens, a bi-prism, and a CMOS camera equipped with a narrow bandpass filter. A fatigue-DIC synchronization technique and an automatic crack tip localization algorithm for numerous high-temperature DIC images are developed, which performs well even under crack branching conditions. Experimental results demonstrate that the automatic localization algorithm achieves an error within 5 % and a computational speed within 3 s per image, effectively addressing the aforementioned challenges and improving analysis efficiency. A new parameter named crack opening ratio (COR) is then applied for the crack closure effect characterization. The evolution laws of the crack closure effect throughout the whole hightemperature fatigue process for pure substrate, pure deposited, and LR specimens are comparatively analyzed, and the influence of the LR interface on the crack growth behavior is clarified. Furthermore, as one major fatigue crack growth model, Paris' law is modified by considering the crack closure effect.