João M.A. Rebello

Bio: João M.A. Rebello is an academic researcher from Federal University of Rio de Janeiro. The author has contributed to research in topics: Ultrasonic testing & Residual stress. The author has an hindex of 19, co-authored 73 publications receiving 1172 citations.

MFL signals and artificial neural networks applied to detection and classification of pipe weld defects

Abstract: This work evaluates the use of artificial neural networks (ANNs) for pattern recognition of magnetic flux leakage (MFL) signals in weld joints of pipelines obtained by intelligent pig. Initially the ANNs were used to distinguish the pattern signals with non-defect (ND) and signals with defects (D) along of the weld bead. In the next step the ANNs were applied to classify signal patterns with three types of defects in the weld joint: external corrosion (EC), internal corrosion (IC) and lack of penetration (LP). The defects were intentionally inserted in the weld bead of a pipeline of API 5L-X65 steel with an outer diameter of 304.8 mm. In this way, the MFL signal itself, digitized with 1025 points, was used as the ANN input. Initially the signals were used as inputs for the neural network without any type of pre-processing, later four types of pre-processing were applied to the signals: Fourier analysis, Moving-average filter, Wavelet analysis and Savitzky–Golay filter. Signal processing techniques were employed to improve the performance of the neural networks in distinguishing between the defect classes. The results showed that it is possible to classify signals of classes D and ND using ANN with very efficient results (94.2%), as well as for corrosion (CO) and LP signals (92.5%). Also it is possible to classify the defect pattern signals: EC, IC and LP using neural networks with an average rate of success of 71.7% for the validation set.

Development of a magnetic sensor for detection and sizing of internal pipeline corrosion defects

Abstract: Magnetic flux leakage (MFL) is the most used technique for pipeline inspection, being applied through the use of instrumented PIGs. The pipe wall is magnetized and when metal loss or other irregularities occur, a larger fraction of the magnetic flux “leaks” outwards from the wall and is detected by sensors. MFL presents some limitations since it requires magnetic saturation of the pipe wall. Therefore, it is difficult to inspect small diameter and thick wall pipelines. Internal corrosion sensor (ICS) has been developed as a solution for internal corrosion measurements of thick walls. The technique, also called “field disturbance”, is based in a direct magnetic response from a small area of the wall. It is not necessary to achieve the magnetic saturation of the pipe material, and thus ICS performance is not affected by the thickness of the pipe wall. In the present work, finite element calculations are performed and the best resultant configuration of the sensor is proposed. Experimental tests with a prototype were carried out and the results give a strong indication of the validity of the theoretical model proposed for sizing.

Pattern recognition of weld defects detected by radiographic test

Abstract: In recent years there has been a marked advance in the research for the development of an automatized system to analyze weld defects detected by radiographs. This work describes a study of nonlinear pattern classifiers, implemented by artificial neural networks, to classify weld defects existent in radiographic weld beads, aiming principally to increase the percentage of defect recognition success obtained with the linear classifiers. Radiographic patterns from International Institute of Welding (IIW) were used. Geometric features of defect classes were used as input data of the classifiers. Using a novel approach for this area of research, a criterion of neural relevance was applied to evaluate the discrimination capacity of the classes studied by the features used, aiming to prove that the quality of the features is more important than the quantity of features used. Well known for other applications, but still not exploited in weld defect recognition, the analytical techniques of the principal nonlinear discrimination components, also developed by neural networks, are presented to show the classification problem in two dimensions, as well as evaluating the classification performance obtained with these techniques. The results prove the efficiency of the techniques for the data used.

Statistical Pattern Recognition

Abstract: This chapter introduces the subject of statistical pattern recognition (SPR). It starts by considering how features are defined and emphasizes that the nearest neighbor algorithm achieves error rates comparable with those of an ideal Bayes’ classifier. The concepts of an optimal number of features, representativeness of the training data, and the need to avoid overfitting to the training data are stressed. The chapter shows that methods such as the support vector machine and artificial neural networks are subject to these same training limitations, although each has its advantages. For neural networks, the multilayer perceptron architecture and back-propagation algorithm are described. The chapter distinguishes between supervised and unsupervised learning, demonstrating the advantages of the latter and showing how methods such as clustering and principal components analysis fit into the SPR framework. The chapter also defines the receiver operating characteristic, which allows an optimum balance between false positives and false negatives to be achieved.

A Generic Deep-Learning-Based Approach for Automated Surface Inspection

Abstract: Automated surface inspection (ASI) is a challenging task in industry, as collecting training dataset is usually costly and related methods are highly dataset-dependent. In this paper, a generic approach that requires small training data for ASI is proposed. First, this approach builds classifier on the features of image patches, where the features are transferred from a pretrained deep learning network. Next, pixel-wise prediction is obtained by convolving the trained classifier over input image. An experiment on three public and one industrial data set is carried out. The experiment involves two tasks: 1) image classification and 2) defect segmentation. The results of proposed algorithm are compared against several best benchmarks in literature. In the classification tasks, the proposed method improves accuracy by 0.66%–25.50%. In the segmentation tasks, the proposed method reduces error escape rates by 6.00%–19.00% in three defect types and improves accuracies by 2.29%–9.86% in all seven defect types. In addition, the proposed method achieves 0.0% error escape rate in the segmentation task of industrial data.