Search Results (58)

Real-time quality monitoring using molten pool images is a critical focus in researching high-quality, intelligent automated welding. To address interference problems in molten pool images under complex welding scenarios (e.g., reflected laser spots from spatter misclassified as porosity defects) and the limited interpretability of deep learning models, this paper proposes a multi-granularity spatiotemporal representation learning algorithm based on the hybrid enhancement of handcrafted and deep learning features. A MobileNetV2 backbone network integrated with a Temporal Shift Module (TSM) is designed to progressively capture the short-term dynamic features of the molten pool and integrate temporal information across both low-level and high-level features. A multi-granularity attention-based feature aggregation module is developed to select key interference-free frames using cross-frame attention, generate multi-granularity features via grouped pooling, and apply the Convolutional Block Attention Module (CBAM) at each granularity level. Finally, these multi-granularity spatiotemporal features are adaptively fused. Meanwhile, an independent branch utilizes the Histogram of Oriented Gradient (HOG) and Scale-Invariant Feature Transform (SIFT) features to extract long-term spatial structural information from historical edge images, enhancing the model’s interpretability. The proposed method achieves an accuracy of 99.187% on a self-constructed dataset. Additionally, it attains a real-time inference speed of 20.983 ms per sample on a hardware platform equipped with an Intel i9-12900H CPU and an RTX 3060 GPU, thus effectively balancing accuracy, speed, and interpretability. Full article

Automated and robotic welding have become standard practices in manufacturing, requiring precise control to maintain weld quality without relying on skilled welders. In Tungsten Inert Gas (TIG) welding, monitoring the weld pool is crucial for ensuring the necessary weld penetration, which is vital for maintaining weld integrity. Real-time observation is essential to prevent defects and improve weld quality. Various sensing technologies have been developed to address this need, with vision-based systems showing particular effectiveness in enhancing welding quality and productivity within the framework of Industry 4.0. This review looks at the latest technologies for monitoring weld pools and bead shapes. It covers methods like using Complementary Metal-Oxide Semiconductors (CMOS) to take clear images of the melt pool for better process identification, Active Appearance Model (AAM) to capture 3D images of the weld pool for accurate penetration measurement, and Charge-Coupled Devices (CCD) and Laser-Induced Breakdown Spectroscopy (LIBS) to analyze plasma spectra and create material composition graphs. Full article

Defects formed during the thermite welding process of two sections of rails require the welded joints to be inspected for quality, and the most used non-destructive method for inspection is radiography testing. However, the conventional defect investigation process from the obtained radiography images is costly, lengthy, and subjective as it is conducted manually by trained experts. Additionally, it has been shown that most rail breaks occur due to a crack initiated from the weld joint defect that was either misclassified or undetected. To improve the condition monitoring of rails, the railway industry requires an automated defect investigation system capable of detecting and classifying defects automatically. Therefore, this work proposes a method based on image processing and machine learning techniques for the automated investigation of defects. Histogram Equalization methods are first applied to improve image quality. Then, the extraction of the weld joint from the image background is achieved using the Chan–Vese Active Contour Model. A comparative investigation is carried out between Deep Convolution Neural Networks, Local Binary Pattern extractors, and Bag of Visual Words methods (with the Speeded-Up Robust Features extractor) for extracting features in weld joint images. Classification of features extracted by local feature extractors is achieved using Support Vector Machines, K-Nearest Neighbor, and Naive Bayes classifiers. The highest classification accuracy of 95% is achieved by the Deep Convolution Neural Network model. A Graphical User Interface is provided for the onsite investigation of defects. Full article

Spatters in the powder-based metal additive manufacturing processes influence deposition quality, part surface quality, and internal defects. We developed a novel video analysis method to monitor and count the melted spatter particles of biomedical-grade Co-Cr-Mo-4Ti powder particles in depositing layers using a micro-plasma transferred arc additive manufacturing (M-PTAAM) process. We captured the spatters using a weld-monitoring camera and building datasets of videos and monitored different combinations of M-PTAAM process parameters. We captured videos of the melted spatter particles and counted the melted spatter particles in real time using a Kalman filter. The weld-monitoring camera captured the melted spatter particles and the fumes generated by the evaporated spatter particles. The video processing algorithm was developed in this study to accurately capture melted spatter particles. In images without fumes, nearly all melted spatter particles were successfully detected. Even in images with the presence of fumes, the algorithm maintained a detection accuracy of 90%. The real-time melted spatter count particle exhibited a powder feed rate changing from 30 to 35 g/min and then to 50 g/min. The melted spatter particle count was lowest at a powder feed rate of 30 g/min and increased with the increasing powder feed rate. Full article

Surface roughness plays a vital role in determining surface integrity and function. Surface irregularities or reduced quality near the surface can contribute to material failure. Surface roughness is considered a crucial factor in estimating the fatigue life of structures welded by FSW. This study attempts to provide a deeper understanding of the nature of the surface formation and roughness of aluminum joints during FSW processes. In order to form more efficient joints, the frictional temperature generated was monitored until reaching 450 °C, where the transverse movement of the tool and the joint welding began. Hardness and tensile tests showed that the formed joints were good, which paved the way for more reliable surface roughness measurements. The surface roughness of the weld joint was measured along the weld line at three symmetrical levels using welding parameters that included a rotational speed of 1250 rpm, a welding speed of 71 mm/min, and a tilt angle of 1.5°. The average hardness in the stir zone was measured at 64 HV, compared to 50 HV in the base material, indicating a strengthening effect induced by the welding process. In terms of tensile strength, the FSW joint exhibited a maximum force of 2.759 kN. Average roughness (Rz), arithmetic center roughness (Ra), and maximum peak-to-valley height (Rt) were measured. The results showed that along the weld line and at all levels, the roughness coefficients (Rz, Ra, and Rt) gradually increased from the beginning of the weld line to its end. The roughness Rz varies from start to finish, ranging between 9.84 μm and 16.87 μm on the RS and 8.77 μm and 13.98 μm on the AS, leveling off slightly toward the end as the heat input stabilizes. The obtained surface roughness and mechanical properties can give an in-depth understanding of the joint surface forming and increase the ability to overcome cracks and defects. Consequently, this approach, using adaptive thresholding image processing coupled with grayscale histogram analysis, yielded significant understanding of the FSW joint’s surface texture. Full article

While the monitoring techniques employed are well established in other fields, their application to the novel processes of ultrasonic and resistance welding of composites is innovative. This study details the adaptation of these techniques to monitor the essential parameters that influence the quality of welds in composite materials within the context of the MFFD (Multi-Functional Fuselage Demonstrator) during the WELDER project. This article presents findings from the application of a sophisticated monitoring system to new welding processes for composites, demonstrating significant enhancements in process control, quality assurance, and operational safety. In ultrasonic welding, temperature, speed, power, amplitude, and displacement were monitored, whereas in resistance welding, voltage, load, and temperature were tracked. This comprehensive data collection is crucial due to the sensitivity of composite materials to welding parameters, which directly affect the integrity and performance of the final product. The data transmission utilized the OPC UA protocol, ensuring secure, reliable real-time data flow to visualization interfaces in safe environments. The integration of real-time monitoring and advanced data analysis helps in identifying potential defects during the welding process, thereby enhancing both the reliability and efficiency of composite welding. The results underscore the potential of these technologies to advance the manufacturing practices for high-performance composite materials in the aeronautics field. Full article

This paper presents an enhanced Faster R-CNN model for detecting surface defects in resistance welding spots, improving both efficiency and accuracy for body-in-white quality monitoring. Key innovations include using high-confidence anchor boxes from the RPN network to locate welding spots, using the SmoothL1 loss function, and applying Fast R-CNN to classify detected defects. Additionally, a new pruning model is introduced, reducing unnecessary layers and parameters in the neural network, leading to faster processing times without sacrificing accuracy. Tests show that the model achieves over 90% accuracy and recall, processing each image in about 15 ms, meeting industrial requirements for welding spot inspection. Full article

Laser welding, widely used in industries such as automotive and aerospace, requires precise monitoring to ensure defect-free welds, especially when joining dissimilar metallic thin foils. This study investigates the application of machine learning techniques for defect detection in laser welding using photodiode signal patterns. Supervised models, including Support Vector Machine (SVM), k-Nearest Neighbors (kNN), and Random Forest (RF), were employed to classify weld defects into sound welds (SW), lack of connection (LoC), and over-penetration (OP). SVM achieved the highest accuracy (95.2%) during training, while RF demonstrated superior generalization with 83% accuracy on validation data. The study also proposed an unsupervised learning method using a wavelet scattering one-dimensional convolutional autoencoder (1D-CAE) network for anomaly detection. The proposed network demonstrated its effectiveness in achieving accuracies of 93.3% and 87.5% on training and validation datasets, respectively. Furthermore, distinct signal patterns associated with SW, OP, and LoC were identified, highlighting the ability of photodiode signals to capture welding dynamics. These findings demonstrate the effectiveness of combining supervised and unsupervised methods for laser weld defect detection, paving the way for robust, real-time quality monitoring systems in manufacturing. The results indicated that unsupervised learning could offer significant advantages in identifying anomalies and reducing manufacturing costs. Full article

Gas metal arc welding (GMAW) is widely used for its productivity and ease of automation across various industries. However, certain tasks in shipbuilding and heavy industry still require manual welding, where quality depends heavily on operator skill. Defects in manual welding often necessitate costly rework, reducing productivity. Vision sensing has become essential in automated welding, capturing dynamic changes in the molten pool and arc length for real-time defect insights. Laser vision sensors are particularly valuable for their high-precision bead profile data; however, most current models require offline inspection, limiting real-time application. This study proposes a deep learning-based system for the real-time monitoring of both the molten pool and weld-bead profile during GMAW. The system integrates an optimized optical design to reduce arc light interference, enabling the continuous acquisition of both molten pool images and 3D bead profiles. Experimental results demonstrate that the molten pool classification models achieved accuracies of 99.76% with ResNet50 and 99.02% with MobileNetV4, fulfilling real-time requirements with inference times of 6.53 ms and 9.06 ms, respectively. By combining 2D and 3D data through a semantic segmentation algorithm, the system enables the accurate, real-time extraction of weld-bead geometry, offering comprehensive weld quality monitoring that satisfies the performance demands of real-time industrial applications. Full article

The fatigue crack growth properties of G20Mn5QT cast steel and corresponding butt welds, using compact tension specimens, were monitored and investigated via acoustic emission (AE) techniques. Fatigue crack growth is a combination of cyclic plastic deformations before the crack tip, tensile crack fractures, and shear crack fractures. The cyclic plastic deformations release the maximum amount of energy, which accounts for half of the total energy, and the second-largest number of AE signals, which are of the continuous-wave type. The tensile crack fractures release the second-largest amount of energy and the largest number of AE signals, which are of the burst-wave type. The shear crack fractures release the least amount of energy and the lowest number of AE signals, which are similar to the burst type, albeit with a relatively longer rise time and duration. Crack tip advancement can be regarded as a discontinuous process. The critical area before the crack tip brittlely ruptures when the fatigue damage caused by cyclic plastic deformations reaches critical status. The ruptures produce a large number of tensile crack fractures and rare shear crack fractures. Through fractography observation, the shear crack fractures occur probabilistically around defects caused by casting or welding, which lead to stress and strain in the local complex. Full article

Real-time quality monitoring through molten pool images is a critical focus in researching high-quality, intelligent automated welding. However, challenges such as the dynamic nature of the molten pool, changes in camera perspective, and variations in pool shape make defect detection using single-frame images difficult. We propose a multi-scale fusion method for defect monitoring based on molten pool videos to address these issues. This method analyzes the temporal changes in light spots on the molten pool surface, transferring features between frames to capture dynamic behavior. Our approach employs multi-scale feature fusion using row and column convolutions along with a gated fusion module to accommodate variations in pool size and position, enabling the detection of light spot changes of different sizes and directions from coarse to fine. Additionally, incorporating mixed attention with row and column features enables the model to capture the characteristics of the molten pool more efficiently. Our method achieves an accuracy of 97.416% on a molten pool video dataset, with a processing time of 16 ms per sample. Experimental results on the UCF101-24 and JHMDB datasets also demonstrate the method’s generalization capability. Full article

Wire-arc additive manufacturing (WAAM) is favored by the industry for its high material utilization rate and low cost. However, wire-arc additive manufacturing of lattice structures faces problems with forming accuracy such as broken rod and surface morphology defects, which cannot meet the industrial demand. This article innovatively combines the melt pool stress theory with visual perception algorithms to visually study the force balance of the near-suspended melt pool to predict the state of the melt pool. First, the method for melt pool segmentation was studied. The results show that the optimized U-net achieved high accuracy in melt pool segmentation tasks, with accuracies of 98.18%, MIOU 96.64%, and Recall 98.34%. In addition, a method for estimating melt pool force balance and predicting normal, sagging, and collapsing states of the melt pool is proposed. By combining experimental testing with computer vision technology, an analysis of the force balance of the melt pool during the inclined rod forming process was conducted, showing a prediction rate as high as 90% for the testing set. By using this method, monitoring and predicting the state of the melt pool is achieved, preemptively avoiding issues of broken rods during the printing process. This approach can effectively assist in adjusting process parameters and improving welding quality. The application of this method will further promote the development of intelligent unmanned WAAM and provide some references for the development of artificial intelligence monitoring systems in the manufacturing field. Full article

Ensuring high-quality control in laser additive manufacturing and laser welding relies on the implementation of reliable and cost-effective real-time observation techniques. Real-time monitoring techniques play an important role in understanding critical physical phenomena, namely, melt pool dynamics and defect formation, during the manufacturing of components. This review aims to explore the integration of acoustic measurement techniques with X-ray imaging for studying these physical phenomena in laser manufacturing. A key aspect emphasized in this work is the importance of time synchronization for real-time observation using multiple sensors. X-ray imaging has proven to be a powerful tool for observing the dynamics of the melt pools and the formation of defects in real time. However, X-ray imaging has limitations in terms of accessibility which can be overcome through combination with other more-accessible measurement methods, such as acoustic emission spectroscopy. Furthermore, this combination simplifies the interpretation of acoustic data, which can be complex in its own right. This combined approach, which has evolved in recent years, presents a promising strategy for understanding acoustic emission signals during laser processing. This work provides a comprehensive review of existing research efforts in this area. Full article

The welding industry plays a fundamental role in manufacturing. Ensuringweld quality is critical when safety, reliability, performance, and the associated cost are taken into account. A ungsten inert gas (TIG) weld quality assessment can be a laborious and time-consuming process. The current state of the art is quite simple, with a person continuously monitoring the procedure. However, this approach has some limitations. Operator decisions can be subjective, and fatigue can affect their observations, leading to inaccuracies in the assessment. In this research project, a deep learning approach is proposed to classify weld defects using convolutional neural networks (CNNs) to automate the process. The dataset used for this project is sourced from Kaggle, provided by Bacioiu et al. The proposed CNN-based approach aims to accurately classify weld defects using the image data. This study trains the model on the welding dataset, using five convolutional layers followed by five pooling layers and, finally, three fully connected layers. The softmax activation function is employed in the output layer to categorize the input into the six weld categories. The per-class metrics, such as precision, recall, and F1-score, suggest that the model is dependable and accurate. Full article

Having garnered significant attention in the scientific community over the past decade, wire-arc directed energy deposition (arc-DED) technology is at the heart of this investigation into additive manufacturing parameters. Singularly focused on Invar as the selected material, the primary objective revolves around devising a virtual sensor for the indirect size measurement of the bead. This innovative methodology involves the seamless integration of internal signals and sensors, enabling the derivation of crucial measurements sans the requirement for direct physical interaction or conventional measurement methodologies. The internal signals recorded, the comprising voltage, the current, the energy from the welding heat source generator, the wire feed speed from the feeding system, the traverse speed from the machine axes, and the temperature from a pyrometer located in the head were all captured through the control of the machine specially dedicated to the arc-DED process during a phase of optimizing and modeling the bead geometry. Finally, a feedforward neural network (FNN), also known as a multi-layer perceptron (MLP), is designed, with the internal signals serving as the input and the height and width of the bead constituting the output. Remarkably cost-effective, this solution circumvents the need for intricate measurements and significantly contributes to the proper layer-by-layer growth process. Furthermore, a neural network model is implemented with a test loss of 0.144 and a test accuracy of 1.0 in order to predict weld bead geometry based on process parameters, thus offering a promising approach for real-time monitoring and defect detection. Full article