Weld Defect Detection Research Articles

Artificial intelligence based defect classification for weld joints

This paper mainly deals with the development of a defect classification system that uses Artificial Neural Network (ANN) to classify weld defects based on ultrasonic test data. The system enables real-time identification of weld defects which finds application in testing of critical welding applications and also reduces dependency on skilled workforce for the function. The study mainly consists of three parts- (i) Weld defect detection using Ultrasonic Testing (UT) (ii) Implementation of ANN (iii) Defect classification. An ultrasonic test performed on welded samples shows different results for welds with and without defects and further between defects as well. The ultrasonic test data is fed into the ANN algorithm to train it to identify the various weld defects. An Artificial Neural Network (ANN) is an information processing paradigm that uses a large number of highly interconnected processing elements called neurons, working in unison to solve the specific problems. There are two types of neural network architectures that are used for classification - a back propagation network (BPN) and a probabilistic neural network (PNN). Back propagation network has been used for the purpose of this study. In order to test the performance of the back propagation neural network, four classes of defect namely porosity, lack of side wall fusion, lack of penetration and slag inclusion are considered.

Modeling for detecting weld defects based on magneto-optical imaging.

A magneto-optical (MO) imaging nondestructive testing (NDT) method for ferromagnetic weldments has been proposed. The mechanism of MO imaging was analyzed by the Faraday MO effect, magnetic domain theory, and magnetic hysteresis loops. Then, the relation between MO images and their corresponding excitation voltages was investigated. To explain the MO imaging system, magnetic domain distribution models of various welding states were established. These models are excited by two kinds of magnetic fields. One is the external magnetic field (Hex), and the other is a weldment remanence field (Mr) after Hex is removed. Relations of magnetic field excitation voltages, thickness of the spacer plate, and the corresponding MO images were also researched, which indicates the proposed NDT method can be used to detect incomplete penetration defect. Then, an experiment that uses MO imaging to detect the defects of high-strength steel (HSS) weldment was performed. Experimental results proved this method can detect crack, sag, and incomplete penetration of weldment effectively. Finally, a series of welded joint MO images of the HSS weldment were captured, which are used as the input data of the defect classification model established by using principal component analysis and an error backpropagation neural network, and the accuracy of this classification model can achieve 92.8%.

An Non Destructive Test for the Detection of Weld Defects Using Image Processing

Welding is a fabrication of joining materials into one component. Defects are unavoidable during the welding process, and hence the inspection of welds is a most important task in many industries. In this work, a Computer Aided Detection (CAD) system is designed to detect weld defects based on image processing techniques. It is a non-destructive testing which uses X-ray images. The proposed system mainly consists of three stages; gradient image formation, filtration by Gaussian pyramidal filters algorithm and segmentation by Expectation and Maximization (EM) algorithm. In this study, GD X-ray weld image database is used to evaluate the proposed system. The performance analysis of the proposed system is done by measuring the sensitivity, specificity, and accuracy of the segmented image with the help of its corresponding ground truth images.

A Review of Nondestructive Examination Methods for New-building Ships Undergoing Classification Society Survey

Classification societies require ship manufacturers to perform nondestructive examination (NDE) of ship weldments to ensure the welding quality of new-building ships. Ships can contain hundreds of kilometers of weld lines and 100% inspection of all welded connections is not feasible. Hence, a limited number of weldments are specified by rules of classification societies to be inspected on a sampling basis. There is a variation between the rules and guidelines used by different classification societies in terms of both philosophy and implementation which results in significant discrepancy in the prescribed checkpoints, numbers, and their locations. In this article, relevant sections of the rules of mainstream International Association of Classification Societies members are studied and potential ways of improving them are discussed. The authors have endeavored to make this study as comprehensive as much as possible. However, given the challenges of covering every single aspect and variable related to NDE in the classification societies' rules and guidelines reviewed here, the authors can only attempt to cover the key features. 1. Introduction Ship structures are joined with hundreds of kilometers of weld lines. The presence of weld in a structure potentially reduces the structure's fatigue life by means of introducing a discontinuity into the completed weld/parent material joint, and introducing residual stress; this could be further amplified by presences of defects inherent to the welding process. Rules, standards, and guidelines may require manufacturers to carry out certain procedures for enhancing weld's reliability, such as weld toe grinding to enhance weld profile geometry (and hence fatigue improvement), heat treatment to improve welded joint's toughness, and nondestructive examination (NDE) to detect weld defects. performing NDE for finished welds is the best way to find possible defects, and relevant rules require manufacturers to do so. However, only a sample of welds is subject to thorough NDE as it is unfeasible to carry out the same extent of detailed inspection on all ship weldments.

Multiframe Ultrasonic TOFD Weld Inspection Imaging Based on Wavelet Transform and Image Registration

TOFD (time of flight diffraction) is a kind of weld defect detection technology by using ultrasonic diffraction wave signal. Because the diffraction intensity is far less than ultrasonic echo wave intensity, thus, the noise contained in TOFD signal is fairly large, and the formed image is not clear enough. Therefore, it is difficult to determine the size of defects accurately. In this paper, a method of noise reduction of TOFD signal and improving the resolution of the image are discussed based on the combination of wavelet thresholding and image registration. Wavelet multiresolution analysis method is adopted and the A-scan signal is decomposed into different frequency components. We propose a new threshold function to process the wavelet coefficients, which guarantees to denoise while preserving the useful information as much as possible. Setting up the ultrasonic TOFD inspection system and the image data with randomly distributed noise can be obtained via fine shake of the probes during testing. Then, image registration based on maximum correlation and blending is adopted to eliminate the noise in further step. The result shows that the proposed method can achieve denoising, together with resolution enhancement.

Development of an ultrasonic weld inspection system based on image processing and neural networks

Several types of discontinuities and defects may be present on a weld, thus leading to a considerable reduction of its resistance. Therefore, ensuring a high welding quality and reliability has become a matter of key importance for many construction and industrial activities. Among the non-destructive weld testing and inspection techniques, the time-of-flight diffraction (TOFD) arises as a very safe (no ionising radiation), precise, reliable and versatile practice. However, this technique presents a relevant drawback, associated to the appearance of speckle noise that should be addressed. In this regard, this paper presents a new, intelligent and automatic method for weld inspection and analysis, based on TOFD, image processing and neural networks. The developed system is capable of detecting weld defects and imperfections with accuracy, and classify them into different categories.

Weldment Nondestructive Testing Using Magneto-optical Imaging Induced by Alternating Magnetic Field

This paper introduces an innovative Nondestructive testing (NDT) approach by using dynamic magneto-optical imaging (MOI) system to diagnose weld defects. MOI mechanism was explained by Faraday magneto-optical effect and magnetic domain theory. Two Q235 specimen MOI experiments based on excitation of permanent magnet and alternating electromagnet (alternating current driven electromagnet) were performed, thus the feasibility of MOI system for weld defects detection was verified and the relation between alternating magnetic field (AMF) and dynamic MO images was discussed as well. In this research, AMF of welded high-strength steel (HSS) weldment was excited by an alternating electromagnet, and dynamic MO images of HSS seam were acquired for weldment NDT. Finally, a pattern recognition method including three steps was proposed. Dynamic MO images were fused periodically and the features of fused images were extracted by principal component analysis. A classifier based on error back propagation (BP) neural network was established to identify these weld features. It proved that typical weld features such as incomplete penetration, sag, crack and no defect can be classified by the proposed method with an accuracy of 93.5%.

Anisotropic diffusion based denoising on X-radiography images to detect weld defects

This paper proposes a machine vision scheme for denoising, feature space gradient preserving, and detecting weld defects in noisy weld X-radiography images; particularly, for the images that are in low-contrast and contain noises. The detection of small weld defects present on noisy image is extremely difficult in non-destructive testing through machine vision. The presence of high gradient magnitude and the low intensity in the feature space of a noisy image are the main characteristics of weld defects. These characteristics can be considered to refine and obtain noise-free images for detection of weld defects. This study proposes a modified anisotropic diffusion model, which considers a local probability value of gray-level and an adaptive threshold parameter in diffusion coefficient function to adjust the implication of low edge gradient of the feature space from the noisy image. Furthermore, an entropy based stopping criterion has been introduced to terminate the diffusion process. This proposed model is compared with the existing models, and its performance is evaluated through Mean Square Error (MSE), Signal-to-Noise Ratio (SNR), Peak Signal-to-Noise Ratio (PSNR), Entropy (E) and Mean Structural Similarity (MSSIM) measures. Experimental results confirm the reliability of the proposed model.

A study on welding quality inspection system for shell-tube heat exchanger based on machine vision

An automated based-vision quality inspection system for Shell-Tube welding is presented in this paper in order to achieve nondestructive weld defect detection. The vision sensor is developed on the basis of the principle of laser triangulation. First, the composition and working principle of the vision-based quality inspection system are introduced, and meanwhile various defects may occur are described in detail. Then the image processing algorithm, as the most important part of online quality inspection system, is also represented. The image processing algorithm includes two parts: preprocessing and defect detection. In defect detection section, a novel method for determining and describing the position of undercut, which is based on the parameter equation of the circle to represent the position of the undercut, is presented. Last, two experiments are carried out for Shell-Tube heat exchanger and the experiments show that the image algorithm has high precision, strong robustness and can detect the defect accurately.

An improved version of Otsu's method for segmentation of weld defects on X-radiography images

It is highly desirable to increase the accuracy of automatic weld defect detection in X-radiography for non-destructive testing and evaluation. For this purpose, a machine vision method is proposed for weld defect segmentation, using the Least Probability Weighted Background Group (LPWBG), an improved version of Otsu's method. This automatically selects a desired threshold value for segmenting the weld defects by adapting the Weibull distribution. The least non-zero probability value of gray-levels of the whole image has been considered as a weighted parameter of the background group of Otsu's within-class criterion. It has been determined that the resulting threshold value will always lie at the left bottom side of the unimodal distribution, especially when the defects are smaller than the background area. Existing approaches such as Otsu's threshold, valley-emphasis, neighborhood valley-emphasis, and weighted object variance have similarly been tried and compared with the proposed method. Our results establish that the proposed method provides satisfactory segmentation results over the others. The performance of the LPWBG method has been both evaluated and compared, using Misclassification Error measure (ME) with significant results.

X-ray online detection for laser welding T-joint of Al–Li alloy

In order to detect weld defects in laser welding T-joint of Al–Li alloy, a real-time X-ray image system is set up for quality inspection. Experiments on real-time radiography procedure of the weldment are conducted by using this system. Twin fillet welding seam radiographic arrangement is designed according to the structural characteristics of the weldment. The critical parameters including magnification times, focal length, tube current and tube voltage are studied to acquire high quality weld images. Through the theoretical and data analysis, optimum parameters are settled and expected digital images are captured, which is conductive to automatic defect detection.

Acoustic emission method for defect detection and identification in carbon steel welded joints

Detecting welding defects in industrial equipment (welded joints and built-up structures) is a key aspect in evaluating the probability of failure in different situations. Acoustic emission (AE) is an effective non-destructive detecting technique, and can be a promising application for welding defect detection. This work presents a systematic experimental investigation on using AE technique for detecting and classifying different weld defects in carbon steel joint material. Four certified carbon steel samples were used in this study. A defect free control sample was used as the reference and three samples with induced defects, namely slag, porosity and crack. A pencil lead break (PLB) test was used to generate simulated AE sources on one side of the joint whereas the AE sensor was mounted on the other side to capture AE signals. A total of four experimental arrangements were used to investigate the effect of propagating distance (sensor to source distance) on the ability of AE to detect and identify defects in welds. For each of these arrangements, AE features such as peak amplitude, rise time, decay time, duration, and count numbers along with statistical features such as AE energy, root mean square (RMS) were extracted and analysed. Also, frequency analysis using FFT and wavelet transform were investigated for each weld test specimen for all arrangements. The results show that AE energy, peak amplitude and RMS value can be used to automatically detect and identify the presence of a defect in carbon steel welds. It is concluded that AE has a considerable potential in use in welding inspection to assess the overall structural health and identify defects that can significantly reduce the strength and reliability of welded material and consequently reduce the risk of component's failure.

Defects detection on the welded reinforcing steel with self-shielded wires by vibration tests

The aim of this paper is the development and validation of a vibroacustic technique to welding defects detection, especially for welded reinforcing structures. In welded structures subjected to dynamic cyclic loads may appear and propagate fatigue cracks due to local structural damage. These cracks may initiate due to the technological parameters used in welding process, or due to environmental operating conditions. By the means of Finite Element Method (FEM), the natural frequencies and shape modes of more welded steel specimens are determined. The analysis is carried out in undamaged condition as well as damaged one, after artificially induced damages. The experimental measurement of the vibroacustic response is carried out by using a condenser microphone, which is suitable for high-fidelity acoustic measurements in the frequency range of 20 – 20.000 Hz. The vibration responses of the welded specimens, in free-free conditions, are carried out using algorithms based on Fast Fourier Transform (FFT), and Prony's series. The results are compared to modal parameters estimated using FE Analysis.

Automatic visual inspection: An approach with multi-instance learning

One of the industrial applications of computer vision is automatic visual inspection. In the last decade, standard supervised learning methods have been used to detect defects in different kind of products. These methods are trained with a set of images where every image has to be manually segmented and labeled by experts in the application domain. These manual segmentations require a large amount of high quality delineations (on pixels), which can be time consuming and often a difficult task. Multi-instance learning (MIL), in contrast to standard supervised classifiers, avoids this task and can, therefore, be trained with weakly labeled images. In this paper, we propose an approach for the automatic visual inspection that uses MIL for defect detection. The approach has been tested with data from three artificial benchmark datasets and three real-world industrial scenarios: inspection of artificial teeth, weld defect detection and fishbone detection. Results show that the proposed approach can be used with weakly labeled images for defect detection on automatic visual inspection systems. This approach is able to increase the area under the receiver-operating characteristic curve (AUC) up to 6.3% compared with the naïve MIL approach of propagating the bag labels.

Stainless Steel Weld Defect Detection Using Pulsed Inductive Thermography

Stainless steel is widely used in many important industries. Stainless steel weld, which is a weak part of the structure, plays a key role in structure integrity. To evaluate the reliability of the structure, this paper proposed a nondestructive testing technique, i.e., pulsed inductive thermography (PIT), to detect defects that are concealed in stainless steel welds. The theory of stainless steel weld defect detection is described, and then, a series of verified experiments are conducted to detect various stainless steel weld defects. The independent component analysis (ICA) is also utilized to remove the influence of weld inhomogeneity and enhance the defect information. The experiments verify that, combined with ICA, the PIT can picture stainless steel weld defects.

Porosity detection by analyzing arc sound signal acquired during the welding process of gas pipeline steel

Detection of severe defects such as porosity underneath the weld bead during the welding process is vital during installation of a gas pipeline network because such defects might lead to fatigue crack. In this study, the work associated with detection of porosity through analysis of the acquired arc sound is presented. Air-borne acoustic signal was acquired during the metal inert gas welding process on API 5L X70 gas pipeline steel. Then, the acquired signal was analyzed using Hilbert Huang transform (HHT), which uses empirical mode decomposition for the purpose of filtering unrelated-to-damage signal components, and Hilbert spectral analysis to obtain the energy-frequency-distance plot. Results showed a significant energy amplitude pattern at the region where both surface- and subsurface-pores existed. Thus, the application of HHT analysis to the acquired arc sound signal has significantly assisted in identifying hidden information that is related to the existence of defects. This finding would enhance the development of an online welding defect detection system during the welding process.

Phased Array Ultrasonic Testing for Post-Weld and OnLine Detection of Friction Stir Welding Defects

ABSTRACTNondestructive evaluation (NDE) techniques of phased array ultrasonic testing (PAUT) and digital X-ray radiography were employed on friction stir (FS)-welded Aluminum Alloy (AA)-2219-T87 specimens. PAUT intricacies required for scanning of FS-welded specimens with a 10-MHz 32-element transducer are discussed. The time corrected gain (TCG) calibration is required for scanning with an increase in index offset to compensate for decrease in A-Scan signal peak amplitude. Calibration techniques to find small defects with appropriate size tolerances are also established. The NDE technique of digital X-ray radiography is compared to PAUT, where it was found that a calibrated PAUT system is able to discover defects less than 0.2 mm where X-ray radiography could not. Incomplete penetration (IP), wormhole (WH), surface cavity (SC), and internal void (IV) defects are analyzed. Furthermore, an online PAUT system for FSW has been developed and successfully tested. The work provided herein will provide a gateway for an ultimate goal of an automated PAUT online sensing system.

GMR-based eddy current probe for weld seam inspection and its non-scanning detection study

Eddy current testing is one of the most important non-destructive testing methods for welding defects detection. This paper presents the use of a probe consisting of 4 giant magneto-resistive (GMR) sensors to detect weld defects. Information from four measuring points above and on both sides of the weld seam is collected at the same time. By setting the GMR sensors’ sensing axes perpendicular to the direction of the excitation magnetic field, the information collected mainly reflects the change in the eddy current which is caused by defects. Digital demodulation technology is applied to extract the real part and imaginary part of the GMR sensors’ output signals. The variables containing directional information of the magnetic field are introduced. Based on the data from the four GMR (4-GMR) sensors’ output signals, four values, Ran, Mean, Var and k are selected as the feature quantities for defect recognition. Experiments are carried out on weld seams with and without defects, and the detection outputs are given in this paper. The 4-GMR probe is also employed to investigate non-scanning weld defect detection and the four feature quantities (Ran, Mean, Var and k) are studied to evaluate weld quality. The non-scanning weld defect detection is presented. A support vector machine is used to classify and discriminate welds with and without defects. Experiments carried out show that through the method in this paper, the recognition rate is 92% for welds without defects and 90% for welds with defects, with an overall recognition rate of 90.9%, indicating that this method could effectively detect weld defects.

Implementation of Welding Defects Detection and Monitoring in Arc Welding Robots

Implementation of Welding Defects Detection and Monitoring in Arc Welding Robots

Defect Detection in Friction Stir Welding Process Using Signal Information and Fractal Theory

Detection of defects in welding is always a challenging task. In friction stir welding process, most of the defects are subsurface which needs special techniques for its detection. In the present research work, an attempt has been made to identify the defects in friction stir welded samples through the information contained in signals acquired during the process. For this analysis, force signals, tool rotational speed signal and main spindle motor current signatures are analyzed using well known fractal theory. Higuchi's fractal dimension algorithm is implemented to extract signal based information in terms of single valued indicator known as fractal dimension. The accountable variation in the computed fractal dimensions against the signals for defective and defect free cases demonstrates the applicability of fractal theory in detection of sub surface defects in friction stir welding process.