Defective Identification Research Articles

NUS: Noisy-Sample-Removed Undersampling Scheme for Imbalanced Classification and Application to Credit Card Fraud Detection

Since minority samples are substantially less common than majority samples, many industrial applications, such as credit card fraud detection (CCFD) and defective part identification, call for imbalanced classification. The performance of a classifier tends to suffer from the noisy samples in majority or minority classes. This work proposes a new undersampling scheme, called a clustering-based noisy-sample-removed undersampling scheme (NUS) for imbalanced classification. The majority class samples are first clustered. The distance of the majority class sample from the cluster center that is furthest away is used as the radius to build a hypersphere, with each cluster’s center assumed to be a spherical center. We determine the Euclidean distance between the center of a cluster and each minority sample to find whether they are in the hypersphere or not. Afterward, we exclude noisy samples from the minority class. The noisy samples of majority classes are removed by using the same procedure. Second, we propose an NUS, which combines noisy sample removal with undersampling techniques. Finally, to prove the effectiveness of NUS, we integrate NUS with the basic classifiers random forest (RF), decision tree (DT), and logistics regression (LR). We conduct their comparison with seven undersampling, oversampling, and noisy-sample-removed methods. This work performs experiments on 13 public and three real transaction datasets related to e-commerce. The results show that NUS plays a positive role in promoting existing classifiers’ performance.

Condition monitoring for fault diagnosis of railway wheels using recurrence plots and convolutional neural networks (RP-CNN) models

RPThe wheel condition monitoring when the train in operation is significant task to prevent the occurrence of unexpected event. In this study, the piezoelectric sensors were installed on the railway track to collect the dynamic voltage-and-strain signals when the train wheels pressed them. These one-dimensional time series signals were transformed to the two-dimensional Recurrence Plots (RP) images as an input data sets for two deep learning models, Xception and EfficientNet-B7. The binary classification, Normal or Faulty as the diagnostical output to indicate the health state of the train wheels in that time. Five metrics were selected to evaluate the performance of two models, namely Accuracy, Precision, Recall, Miss Rate, and AUC. The results show that both models perform the high accuracy of 91.1% to the wheel condition classification. Furthermore, EfficientNet-B7 shows better performance in Recall, Miss-rate, and AUC metrics than those of Xception to express the premium ability in defective wheel identification, which is crucial for this application. Therefore, the efficientNet-B7 is selected as a favorable machine learning classifier for the fault diagnosis of rolling stock wheels. It is significant contribution to train wheel condition monitoring and health management since it provides the effective diagnostic information for maintenance decision to decrease the occurrence of unexpected event.

Exploratory Data Analysis for Textile Defect Detection

The capacity to recognize anomalies in real-world visual data is essential for many computer vision uses. New approaches and ideas in unsupervised defective garments identification require data for training and evaluation. Understanding the constraints of the currently employed approach of human inspection is crucial for improving clothing quality. Uses for digital image processing in the textile sector are suggested. This method proposes a novel quantitative measuring strategy by fusing digital image processing with the Lab view platform. As this study progresses, it becomes clear that the FLDA yields the best results, with 95% accuracy, while the Hoeffiding Tree yields the lowest results, with 60% accuracy. When compared to other models, the FLDA's precision of 0.96 is the best you'll find, while the Hoeffiding Tree's is the lowest at 0.62. The FLDA provides the best result, with a recall value of 0.95, while the Hoeffiding Tree shows the lowest result, with a recall value of 0.60. The FLDA yields the best results (0.90 kappa value), whereas the Hoeffiding Tree yields the worst (0.20 kappa value).The FLDA exhibits the best results, with an F-Measure value of 0.95, while the Hoeffiding Tree displays the lowest results, with an F-Measure value of 0.58. The FLDA provides the best results, with an MCC value of 0.91, while the Hoeffiding Tree displays the worst results, with an MCC value of 0.22. The FLDA yields the best results (0.98 ROC value), whereas the Decision Table produces the worst results (0.69 ROC value). The best prediction accuracy is shown by the FLDA, at 0.98 of the PRC value, while the worst is shown by the Decision Table, at 0.67. The MAE is lowest (0.07) for the FLDA and highest (0.39) for the Hoeffiding Tree. The MAE deviation of the Bayes Net is 0.19. The best result is shown by the FLDA, with an RMSE of 0.22, while the largest RMSE deviation is found in the Hoeffiding Tree, at 0.62. The RMSEdeviation for Bayes Net is 0.41. The finest RAE is shown by the FLDA, at 13.39%, while the largest RAE deviation is 78.28% for the Hoeffiding Tree. The Bayes Net explains 38.74% of the variation in RAE. The best result is shown by the FLDA, with an RRSE of 44.36%; the largest RRSE variation is shown by the Hoeffiding Tree, with 123.99%. When compared to other models, the IBK's preparation time of 0 seconds is by far the shortest. While the Bayes Net completes its task in 0.03 seconds, FLDA can take up to 0.17 seconds. The FLDA model is found to have superior performance in this study.

Slag and Porosity Defective Region Identification in Welding Images Using Computer Vision Techniques

Slag and Porosity Defective Region Identification in Welding Images Using Computer Vision Techniques

Investigations for Material Tracing in Selective Laser Sintering: Part ΙΙ: Validation of Modified Polymers as Marking Agents

Selective laser sintering (SLS) is currently in transition to the production of functional components. However, the ability to apply it is confronted with new requirements for reliability and reproducibility. Therefore, an in-depth understanding of aging processes in polymers is essential. Regarding material traceability as well as defective component identification with subsequent cause tracing, the application of a material-inherent marking technology represents a solution. SLS in combination with modified polymers as a marking technology proves to be an efficient opportunity to produce reproducible and high-quality components due to an increased understanding of the process. Based on a selection of modified polymers for use in SLS, which were characterized in part I of the study, this work focuses on the experimental validation of the result. The influence of modified polymers on materials and component properties and the SLS process's influence on the traceability of modified polymers are examined. Intrinsic and extrinsic material properties as well as mechanical properties, surface quality and sinter density are analyzed. No discernible influences of the modified polymers on the investigated properties could be observed and the traceability of the modified polymers could also be confirmed in the aged powder and component using mass spectroscopy.

Inline Defective Laser Weld Identification by Processing Thermal Image Sequences with Machine and Deep Learning Techniques

The non-destructive testing methods offer great benefit in detecting and classifying the weld defects. Among these, infrared (IR) thermography stands out in the inspection, characterization, and analysis of the defects from the camera image sequences, particularly with the recent advent of deep learning. However, in IR, the defect classification becomes a cumbersome task because of the exposure to the inconsistent and unbalanced heat source, which requires additional supervision. In light of this, authors present a fully automated system capable of detecting defective welds according to the electrical resistance properties in the inline mode. The welding process is captured by an IR camera that generates a video sequence. A set of features extracted by such video feeds supervised machine learning and deep learning algorithms in order to build an industrial diagnostic framework for weld defect detection. The experimental study validates the aptitude of a customized convolutional neural network architecture to classify the malfunctioning weld joints with mean accuracy of 99% and median f1 score of 73% across five-fold cross validation on our locally acquired real world dataset. The outcome encourages the integration of thermographic-based quality control frameworks in all applications where fast and accurate recognition and safety assurance are crucial industrial requirements across the production line.

Privacy-Preserving KNN Classification Algorithm for Smart Grid

With the development of Internet of Things (IOT), outsourcing data and tasks to a cloud server has become a popular and economical way for small devices with restricted ability. The k-nearest neighbors (KNNs) classification algorithm have been commonly applied to medical image classification, abnormal detection, defective product identification, and so on. The previous privacy-preserving KNN algorithms are based on two cloud servers, which have high computational and communication costs. In this paper, we design a privacy-preserving KNN classification (PPKC) algorithm with single cloud server for smart grid. Specifically, each smart meter and control center encrypted their data with Paillier cryptosystem and the cloud server does some calculation on the encrypted data. We prove that PPKC can protect the privacy of both smart meters and control centers, and the classification results are also private for the server. Besides, both smart meters and control centers can stay offline after uploading their data. The experiment results demonstrate that the proposed PPKC algorithm is more efficient than the previous algorithms and it can obtain almost the same accuracy as the original KNN, which means the PPKC algorithm is more applicable for small devices in IOT.

Designing a digitalized cell for remanufacturing of automotive frames

Circular economy introduction in industry is globally recognized as one of the most effective green strategies towards sustainability and resources reduction. As the automotive sector has been historically one of the biggest energy consumers, significant efforts are being constantly made to reduce the environmental impact and to increase energy efficiency at the manufacturing stage, both at production and process level. This study focuses on proposing a framework for automotive defective parts identification and remanufacturing in the main production line. In particular, the concept of an automotive microfactory is taken into consideration. To achieve this, a vision-based monitoring system performing the defect identification drives the whole procedure that is considered to be realized through a remanufacturing cell. The software tool is considered to be utilized right before the scheduling process of orders and tasks performed in the remanufacturing cell, taking into account specific performance parameters. The architecture of the software is presented along with some discussion regarding the types of control that can be enforced at both processing and line level, while the use of synthetic data is regarded to test the efficiency of the monitoring system.

Defective products identification framework using online reviews

Defective products identification framework using online reviews

Fault diagnosis of rolling element bearing using autonomous harmonic product spectrum method

Maintenance planning plays a critical role in the process industry, where any unplanned maintenance may lead to a significant loss. Condition monitoring happens to aid maintenance planning and has become an inherent part of the maintenance activity. Physical parameters such as vibration, acoustic emission, current, etc., are used for condition monitoring, out of which vibration is the most preferred parameter and is widely used in the industry. Vibration data is measured near to bearings, which themselves are monitored for their condition, and hence rolling element bearing (REB) is the focus of this study. REBs are monitored for the presence of a fault in them as well as for their severity. Fault diagnosis of REB using harmonic product spectrum (HPS) is proposed in this study. The proposed methodology's novelty lies in the signal pre-processing step, whose output is fed to the HPS method, which is used for defective raceway identification. The efficacy of HPS is assessed with parameter optimized Variational mode decomposition (VMD) and classical bandpass filtering method as pre-processors. It is observed that the HPS delivers better diagnostic results with the VMD method than the bandpass filtering method. Non-dominated sorting particle swarm optimization algorithm is deployed for parameter optimization of VMD. HPS combined with VMD as pre-processor forms an autonomous HPS(AHPS) algorithm, whose input is measured signal and output is defect frequency. The process is so designed that a raw signal, when fed to the algorithm, delivers the result as identification of a defective raceway. Unlike previously developed methods, the proposed method needs no manual intervention. Results obtained from simulated signals and signals recorded through experiments validate that the proposed methodology can be used effectively for fault diagnosis of REB.

Predicting Defective Lines Using a Model-Agnostic Technique

Defect prediction models are proposed to help a team prioritize the areas of source code files that need Software Quality Assurance (SQA) based on the likelihood of having defects. However, developers may waste their unnecessary effort on the whole file while only a small fraction of its source code lines are defective. Indeed, we find that as little as 1-3 percent of lines of a file are defective. Hence, in this work, we propose a novel framework (called <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> ) to identify defective lines using a model-agnostic technique, i.e., an Explainable AI technique that provides information why the model makes such a prediction. Broadly speaking, our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> first builds a file-level defect model using code token features. Then, our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> uses a state-of-the-art model-agnostic technique (i.e., LIME) to identify risky tokens, i.e., code tokens that lead the file-level defect model to predict that the file will be defective. Then, the lines that contain risky tokens are predicted as defective lines. Through a case study of 32 releases of nine Java open source systems, our evaluation results show that our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> achieves an average recall of 0.61, a false alarm rate of 0.47, a top 20%LOC recall of 0.27, and an initial false alarm of 16, which are statistically better than six baseline approaches. Our evaluation shows that our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> requires an average computation time of 10 seconds including model construction and defective line identification time. In addition, we find that 63 percent of defective lines that can be identified by our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> are related to common defects (e.g., argument change, condition change). These results suggest that our <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Line-DP</small> can effectively identify defective lines that contain common defects while requiring a smaller amount of inspection effort and a manageable computation cost. The contribution of this paper builds an important step towards line-level defect prediction by leveraging a model-agnostic technique.

Fabric Defect Detection Using the Sensitive Plant Segmentation Algorithm

Fabric quality control and defect detection are playing a crucial role in the textile industry with the development of high customer demand in the fashion market. This work presents fabric defect detection using a sensitive plant segmentation algorithm (SPSA) which, is developed with the sensitive behaviour of the sensitive plant biologically named “mimosa pudica”. This method consists of two stages: The first stage enhances the contrast of the defective fabric image and the second stage segments the fabric defects with the aid of the SPSA. The SPSA proposed was developed for defective pixel identification in non-uniform patterns of fabrics. In this paper, the SPSA was built through checking with devised conditions, correlation and error probability. Every pixel was checked with the algorithm developed to be marked either a defective or non-defective pixel. The SPSA proposed was tested on different types of fabric defect databases, showing a much improved performance over existing methods.

Secure and Efficient k NN Classification for Industrial Internet of Things

The ${k}$ -nearest neighbors ( ${k}$ NN) classification has been widely used for defective product identification and anomaly detection in the Industrial Internet of Things (IIoT). In this article, we propose a secure and efficient distributed ${k}$ NN classification algorithm (SEED- ${k}$ NN) to prevent information and control flow exposure while supporting large-scale data classification on distributed servers. Specifically, we first design a secure and efficient vector homomorphic encryption (VHE) scheme by constructing a key-switching matrix and a noise matrix for data encryption. Based on the designed VHE, SEED- ${k}$ NN is proposed to efficiently achieve the confidentiality of data flow, ${k}$ NN query, and class label, while enabling homomorphic operations on the encrypted data. Moreover, by leveraging the Map/Reduce architecture, SEED- ${k}$ NN enables the ${k}$ NN classification over the large-scale encrypted data on distributed servers for industrial control systems. Finally, we demonstrate that SEED- ${k}$ NN achieves semantic security and high classification accuracy, and is applicable in IIoT due to its high efficiency.

Defective tissue identification from crowded tissue cluster of 3D images

This research work formulates the challenge of defective tissue identification in crowded tissue clusters. An identified set of 3D tissue images are matched with predefined defective tissue images by using image evidence. Image evidence is obtained by foreground extraction and boundary probability. An optimum solution is derived by the decomposition of conjointly related tissues hooked on occluded plus unoccluded ones in every sequence of iteration. Only unoccluded candidates will be the matching models by the tissue relations of graph description. Tissue authentication and elimination processes are built on minimal descriptor dimension and confined occlusion, which is conceded out after each and every sequence of model equivalent. The computational cost of the proposed method is comparatively much smaller than the comprehensive optimization techniques. Tissue Detection rate is identified to be 2% higher than the result provided by previous means. The performance is demonstrated by the proposed method using another set of challenging tissue datasets. The outcome of our proposed technique is the best than the existing OTSU, HMT, 2DORG, and WK methods.

A Flux-Based Differential Technique for Turn-to-Turn Fault Detection and Defective Region Identification in Line-Connected and Inverter-Fed Induction Motors

A new air-gap flux-based technique is proposed in this paper for turn-to-turn fault (TTF) detection and defective region identification, which is implemented in line-connected and inverter-fed induction motors. Some search coils (SCs) are used to measure the magnetic flux in various air-gap regions of the mentioned motor. The induced voltages in the SCs reveal the configuration of the rotational magnetic field (RMF) along the stator inner circumference to assess the symmetricity level of the flux distribution at each time instant. The RMF asymmetric distribution can be detected by comparing the induced voltages in various dual SCs (which under healthy conditions must sense identical magnetic flux) to detect TTFs and identify the faulty region as well. The proposed method is verified through finite element simulations and also by carrying out several experimental tests. Despite the fact that the proposed method is an invasive diagnostic technique, significant sensitivity and robustness in the face of unbalanced voltage sources and/or loading level are compelling evidence to use it for TTF detection at an incipient stage. The capability of the proposed technique to identify the faulty phase and the related region in the stator is another advantage of this invasive method, which cannot be ignored.

Distributed Capacitive Sensors for Partial Discharge Detection and Defective Region Identification in Power Transformers

Analyzing of detected partial discharge (PD) signals is a reliable technique to assess the electrical insulation in power transformers. In order to detect and locate the PD sources in power transformers, some distributed simple fine sensors are proposed and used, which are made by a few turns of thin enameled copper wire wrapped around the transformer core. They are distributed sensors, which sense PD voltage signals through capacitive coupling with the transformer windings. Two methods, including broadband and limited-band PD signal analyzing, are considered to identify defective phase and/or region on that phase through the proposed distributed sensors outputs. Transfer functions of these sensors are obtained and it is experimentally shown that although defective phase can be diagnosed, the expected priority to identify defective region is not regarded in broadband analysis, while a suitable limited frequency band can be found to achieve this aim. The idea is implemented on a sample transformer by injecting artificial PD signals to different high-voltage winding sections. The results show that not only PD signals can be detected through the mentioned sensors with high sensitivity, but also defective phase can be diagnosed using broadband analysis. Moreover, defective region can be accurately identified by limited-band analysis.

Extreme Learning Machine and Moving Least Square Regression Based Solar Panel Vision Inspection

In recent years, learning based machine intelligence has aroused a lot of attention across science and engineering. Particularly in the field of automatic industry inspection, the machine learning based vision inspection plays a more and more important role in defect identification and feature extraction. Through learning from image samples, many features of industry objects, such as shapes, positions, and orientations angles, can be obtained and then can be well utilized to determine whether there is defect or not. However, the robustness and the quickness are not easily achieved in such inspection way. In this work, for solar panel vision inspection, we present an extreme learning machine (ELM) and moving least square regression based approach to identify solder joint defect and detect the panel position. Firstly, histogram peaks distribution (HPD) and fractional calculus are applied for image preprocessing. Then an ELM-based defective solder joints identification is discussed in detail. Finally, moving least square regression (MLSR) algorithm is introduced for solar panel position determination. Experimental results and comparisons show that the proposed ELM and MLSR based inspection method is efficient not only in detection accuracy but also in processing speed.

Electroluminescence of silicon solar cells using a consumer grade digital camera

Abstract Electroluminescent imaging is increasingly used to detect defects in silicon solar cells. However, the cost of the conventional luminescence systems is a limiting factor for generalized use. A simple and reliable low-cost electroluminescence setup is presented. The developed system was tested on commercial silicon solar cells for the acquisition of electroluminescence images in the forward and reverse bias regimes. In forward bias the temperature was varied whilst in reverse bias the applied voltage was varied. The results used in conjunction allowed for the detection of defective areas and identification of their type and cause. The simplicity and low-cost nature of the developed setup should enable a more widespread use of luminescence techniques for the characterization of crystalline silicon solar cells.

Feasibility study for partial discharge detection and defective region identification in power transformers by using distributed sensors

Monitoring of partial discharge (PD) and its progress as the main cause of the insulation deterioration process in power transformers can encompass valuable information about the related insulation condition and reveals degradation mechanisms. In this study, some simple sensors, which are made from a few turns wrapped around the transformer core, are used to detect PD signals. Moreover, a feasibility study is done to identify defective phase/region based on the captured PD signals waveforms through these sensors. The idea is implemented on a 50 kVA, 20 kV/400 V three-phase transformer by injecting artificial and quasi-real PD signals to different sections of HV windings. Also, various transformer connections are considered in the experimental studies to find the effect of them on the waveforms of the captured signals (e.g. peak-to-peak value and signal energy). The obtained experimental results show that small PDs (as low as 10 pC) is detectable by the mentioned sensors and defective phase can be identified in all of transformer connection cases. Furthermore, considering failure to identify defective regions by broadband analysis, a limited-band approach is proposed through obtaining |H(s)| for the sensors. The results show that exact defective phase/region identification is feasible by selecting a frequency band compatible with IEC60270 recommendation.

Defective Sensor Identification for WSNs Involving Generic Local Outlier Detection Tests

The behavior of a wireless sensor network dedicated to distributed estimation tasks may be significantly altered by the presence of nodes whose sensors are defective and produce erroneous measurements. This paper proposes and analyzes the performance of two distributed algorithms to help each node in determining whether it is equipped with a defective sensor. A node first collects data from its neighborhood, processes them to decide, using some generic local outlier detection test, whether these data contain outliers and broadcasts the result. Then, it determines the status of its own sensor using its result and those received from neighboring nodes. A single-decision and an iterative algorithm for defective sensor detection are proposed. Bounds on the performance of the single-decision algorithm are derived. A theoretical analysis of the probability of error and of the equilibrium of the iterative algorithm is provided for a wide class of local outlier detection tests. The tradeoff between false alarm probability and detection probability is characterized theoretically and by simulation. MAC-layer issues, as well as the effect of packet losses are accounted for.