Root Cause Of Fault Research Articles

Deformable Convolutional Networks for Efficient Mixed-Type Wafer Defect Pattern Recognition

Defect pattern recognition (DPR) of wafer maps is critical for determining the root cause of production defects, which can provide insights for the yield improvement in wafer foundries. During wafer fabrication, several types of defects can be coupled together in a piece of wafer, it is called mixed-type defects DPR. To detect mixed-type defects is much more complicated because the combination of defects may vary a lot, from the type of defects, position, angle, number of defects, etc. Deep learning methods have been a good choice for complex pattern recognition problems. In this article, we propose a deformable convolutional network (DC-Net) for mixed-type DPR (MDPR) in which several types of defects are coupled together in a piece of wafer. A deformable convolutional unit is designed to selectively sample from mixed defects, then extract high-quality features from wafer maps. A multi-label output layer is improved with a one-hot encoding mechanism, which decomposes extract mixed features into each basic single defect. The experiment results indicate that the proposed DC-Net model outperforms conventional models and other deep learning models. Further results of the interpretable analysis reveal that the proposed DC-Net can accurately pinpoint the defects areas of wafer maps with noise points, which is beneficial for mixed-type DPR problems.

Surface Defect Detection and Root Cause Analysis

Artificial Intelligence has played an increasingly important role in surface defect detection in recent years. At the same time, there are many challenges using deep learning for this area, such as the detection accuracy, shortage of data and, lack of knowledge of root cause of defects. To solve the problem of data shortage, we propose a taxonomy method called DataonomyTM to extend a meta defect datasets with a small number of samples for training defect classifiers. For the accuracy, we apply two latest deep neural network(DNN) architectures, Inception v3 and fully convolutional networks (FCN) so as not only to classify whether there are defects but also to make a pixel-wise prediction to inference the areas of defects. For those detected defects, we combine DNN with traditional AI methods to find root causes of detected defects. We use a generalized multi-image matting algorithm to extract common defects automatically. We apply this technology to identify defects that stem from systematic errors in the surface operation. Experimental results have shown great capability and versatility of our proposed methods.

Root cause analysis of pinhole defects on painted galvanized steel panel

Galvanized steel sheets have been widely used in automotive industries. However, it is still a challenge for steel makers to produce defect free hot dip galvanized steel sheets especially for outer panel applications in automotive body. In this study, the root cause of uncommon pinhole defects on painted hot dip galvanized steel panel was investigated. Optical microscopy (OM), scanning electron microscope (SEM) coupled with energy dispersive spectrometry (EDS), and auger electron spectroscopy (AES) were used to characterize these pinhole defects from top surface to base steel layer by layer. A reannealing experiment was carried out to reproduce the key features of the corresponding defect on the base steel. It was found that the pinhole defects were composed of a series of connected voids in the paint layers and started from the dents on the base steel prior to hot dip galvanizing. Based on the results of coarse grains around the dents and residual boron oxides at the dent bottom, further cause of the dents was attributed to the boron containing buildups on furnace rolls in the high temperature section of hot dip galvanizing line. As the surface of furnace rolls was cleaned up, the dent defects on the base steel, the bare spot defects in the zinc coating and the pinhole defects on the painted panel were eliminated successfully.

Fault Mechanism of Fiber Optical Current Transformer Based on Signal Processing Method

The fiber optical current transformer (FOCT) with sine wave modulation has been widely used in the DC transmission project, while its fault rate is much higher than that of an electromagnetic current transformer, which seriously threatens the safe operation of power grid. In order to investigate the fault mechanism of FOCT, the output signal model of the FOCT photodetector is established, and its basic methods of modulation and demodulation are analyzed to explore the influencing factors of the signal processing method used in the calculation of measured current. It is found that the reduction of the second harmonic component in the photodetector output is the root cause of FOCT fault, and the optical path, modulation signal amplitude, and measured current are important factors affecting the second harmonic component. The influencing law of three factors on the second harmonic component is calculated by simulation and the characteristic differences between them are given. Finally, the fault simulation test of FOCT light intensity and a modulation circuit were carried out in the laboratory, and the results verify the correctness of theoretical analysis. The research in this paper provides reference for fault analysis and reliability improvement of FOCT.

A Novel Intelligent Fault Diagnosis Approach for Critical Rotating Machinery in the Time-frequency Domain

The rotating machinery is a common class of machinery in the industry. The root cause of faults in the rotating machinery is often faulty rolling element bearings. This paper presents a novel technique using artificial neural network learning for automated diagnosis of localized faults in rolling element bearings. The inputs of this technique are a number of features (harmmean and median), which are extracted from the vibration signals of the test data. Effectiveness and novelty of this proposed method are illustrated by using the experimentally obtained the bearing vibration data based on laboratory application. In this research, based on the fast kurtogram method in the time-frequency domain, a technique for the first time is presented using other types of statistical features instead of the kurtosis. For this study, the problem of four classes for bearing fault detection is studied using various statistical features. This study is conducted in four stages. At first, the stability of each feature for each fault mode is investigated, then resistance to load change as well as failure growth is studied. At the end, the resolution and fault detection for each feature using the comparison with a determined pattern and the coherence rate is calculated. From the above results, the best feature that is both resistant and repeatable to different variations, as well as the suitable accuracy of detection and resolution, is selected and with comparing to the kurtosis feature, it is found that this feature is not in a good condition in compared with other statistical features such as harmmean and median. The results show that the accuracy of the proposed approach is 100% by using the proposed neural network, even though it uses only two features.

Transgenerational effects following chronic circadian disturbance given to mother rats in F1 generation rats

It has not been documented whether circadian disturbance of mother may influence fetal neurodevelopment. Therefore, this study aims to find the root cause of congenital neuronal defects, which is explosively increasing today, in one of the environmental factors; circadian disturbance. Female rats were randomly divided into 3 groups; L/D group with 12/12 hr light and dark cycle, LL group with 24 hr constant light, DD group with 24 hr constant dark. All subjected females were assigned to each light condition one week prior to pregnancy. Then, they were mated with males, and F1 offsprings were obtained at three weeks after the mating. The future generations were exposed exactly same light conditions. The animals in each group were randomly divided into control and melatonin‐treated subgroups. All animals were engaged in physiological, histological, and neurobehavioral analyses. LL and DD groups showed several abnormal features and decreased cecum/colon length (p<.05). Also, LL and DD groups showed lower activation of intracranial δ‐wave than L/D group showed. However, in LL and DD groups treated with melatonin improved intestinal integrity and activated intracranial δ‐wave. Axonal sprouting was evaluated by Golgi‐cox staining for identifying the abnormal neurodevelopment in all groups. The LL and DD groups showed reduction of axonal sprouting than L/D group. However, axonal branching in neuron increased in LL and DD groups treated with melatonin. In addition, neurobehavioral tests including open field test, novel objective recognition test, social interaction test, spontaneous alternation Y‐maze test, and elevated plus maze test were performed. LL and DD groups showed abnormal neurobehaviors (p<.05), whereas, melatonin treatment resulted in the normalized neurobehaviors (p<.05). In conclusion, F1 generation born from mothers exposed to chronic circadian disturbance showed abnormal neurodevelopment and neurobehaviors. Melatonin was considered as a therapeutic agent for congenital neuronal deficit.Support or Funding InformationFundings: National Research Foundation (NRF‐2017R1A2A2A01067169, NRF‐2019R1A6A3A01091422), Korea

Image Defect Detection and Root Cause Analysis

Artificial Intelligence has played an increasingly important role in visual defect detection in recent years, while there are many challenges using deep learning for this application, such as the shortage of data, lack of knowledge of root cause of defects. In this paper, we combine deep learning with traditional AI methods, not only to solve unshaded defect detection but also find root causes of detected defects. First, we propose a taxonomy method called DataonomySM to extend a meta defect dataset with a small number of samples and a deep learning method to detect the image defects. For detected defect images, we use a generalized multi-image matting algorithm to extract common defects automatically. We apply this technology to identify defects that stem from systematic errors in a product line and later extended its use to watermark processing. Experimental results have shown great capability and versatility of our proposed methods.

Автоматизація обробки аварійних подій в інформаційно-технологічній інфраструктурі організації

Telecommunication and internet providers manage IT infrastructures consisting of tens of thousands of various devices. Monitoring and managing such infrastructure is quite a complicated task demanding pro-fessional and coherent job from technical support teams. One of key tasks for monitoring service is prompt detection of equipment faults and quick engagement of corresponding experts to restore normal operation of equipment and systems. Corresponding tools are required to automate this process, which is quite la-bour-consuming. Orion CRL is offered as a tool with such functionality. It is integrated with the systems for monitoring, incident management, network management, inventories, CRM, Service Desk, IVR, contact center. It accelerates addressing critical faults giving additional opportunities for basic diagnostics. False diagnostics or mistaken incident registration leads to time losses, delays elimination of problems and dis-tracts specialist from handling other urgent issues and tasks. To increase diagnostics efficiency there has been developed an original algorithm of analysis and search for root cause of fault; so called Root Cause Analysis (RCA). Algorithm is based on network topology analysis and faults which are detected by moni-toring system such as Host Down, Host Up, Link Down. Algorithm provides for basic diagnostics, typical fault check operations on restore procedures which are typical for organization's IT infrastructure. Due to integration with monitoring system the RCA discovers root cause and symptomatic events. At the same time Orion CRL disguises faults that are symptomatic and highlights the root cause in monitoring system. The developed Orion CRL tool set increases efficiency of monitoring service. Automation of typical basic diag-nostics operations and fault detection, in its turn, accelerates restoration of normal operation of equipment and saves operators’ work time.

Deep learning enhanced digital twin for Closed-Loop In-Process quality improvement

A digital twin framework is presented for assembly systems with compliant parts fusing sensors with deep learning and CAE simulations. Its underlying concept, ‘process capability space,’ updates iteratively during evolving tasks of new product introduction with resulting model fidelity able to simulate dimensional, geometric and weld quality of parts and assemblies; isolate root causes of quality defects; and, suggest corrective actions for automatic defects mitigation; thereby, enabling ‘Closed-Loop In-Process (CLIP) quality improvement’ during assembly system development. Results, using the first fully digitally developed remote laser welding process for aluminium doors, yielded a right-first-time rate of >96% for door assembly cell development.

Root causality analysis at early abnormal stage using principal component analysis and multivariate Granger causality

As fault detection technologies have been developed, process fault diagnosis at early abnormal stage has come to be considered a major problem. In this work, a method to analyze the root cause of faults is developed to provide proper information at the early abnormal stage. First, principal component analysis (PCA) is used for the early detection of the process fault. Then, the contributions, from which the normal portion is removed, are decomposed by singular value decomposition (SVD) method to select the hierarchical sensors. Finally, the multivariate Granger causality (MVGC) method is used to construct the sensor causalities using the hierarchical sensors. The developed methodology is verified using the liquefied natural gas fractionation process model, which embeds a sufficient number of highly correlated sensors. The results are compared with the conventional principal component analysis method and amplification of the residual contribution method to verify the advantages of the proposed method.

Investigation of defects influencing performance of type-II InAs/GaInSb superlattice based infrared PIN type photodetectors

We report on an investigation of dark current contributions from common microscale crystallographic defects in type-II InAs/GaInSb superlattice infrared PIN type photodiode structures grown on (100) GaSb substrates and have identified three general classifications. Defects on several wafers of varying design were examined from multiple perspectives to correlate electrical activity with structural properties, to develop the identification and classification scheme. Active defects were first identified by current density vs voltage (J-V) measurements and electron beam induced current (EBIC) scans of individual diodes with micrometer resolution. The EBIC scans were then correlated with plan-view optical and atomic force microscopy images, both before and after anisotropic etch-pit analysis using a newly developed etchant. The atomic scale structure of active and inactive defects was then compared using cross-sectional transmission electron microscopy (TEM) on vertical slices of defects extracted using focused ion beam milling. Analysis of the TEM images yielded important clues as to the structure and root causes of benign and active defects, in which only significant disruptions at the epi-substrate interface appear to play a key role in producing microscale defects that efficiently promote dark current.

Kaizen a Tool for Continuous Quality Improvement in Indian Manufacturing Organization

Kaizen is a practice of continuous improvements in manufacturing, assembly, quality and engineering. In this paper author have discussed the problems faced by a leading automobile manufacturing organization on quality of interior parts of the automobile like plastics and rubber. During study author used a new technique called “Global Customer Audit” which is basically a vehicle audit from final customer point of view for interior parts of automobiles. During this audit, author has observed various defects and by using different quality control tools like pareto charts, histograms, check sheets and arranged defects according to their intensity or magnitude of occurrence i.e. finding useful one from trivial many. In this paper, the author found the root cause of defects by using different quality control tools. Kaizen is done to remove these defects and continuously improve products and production processes. Kaizen helps in making manufacturing process leaner, simple and fitter. The main purpose of this paper is to discuss the concept of Kaizen, its application in case organization for solving problems raising big quality concerns on a day to day bases.

Lateral Forces in Rolling-Cut Shearing and Their Consequences on Common Edge Defects

Abstract This paper deals with the detailed analysis of the lateral process forces in rolling-cut shearing of heavy steel plates and their impact on edge defects. Rolling-cut shearing is still the most common method of heavy-plate side trimming. However, this method can entail edge defects like uneven longitudinal shape as well as burr and fractures in the area of the cut-changeover (beginning and end of the periodical cuts). In the existing literature, neither the root cause of these edge defects nor their nexus with the upper blade trajectory (blade drive-kinematics) has been analyzed in detail. In this work, these issues will be explored based on the finite element method (FEM) simulations and measurements from an industrial plant. The complex interrelation between drive-kinematics, varying lateral force, unintended lateral motion of the upper blade, unintended variation of the blade clearance, and quality defects is analyzed. The variation of the lateral force is identified as the root cause of such quality defects and a physical explanation for variations of the lateral force is given. The detailed understanding of the shearing process serves as a solid basis for an optimization and re-design of the drive-kinematics in a future work. Measurements from an industrial plant and simulation results show good agreement and thus confirm the theory. The results are transferable to other rolling-cut trimming shears.

A Case Study on Defining a “Quality Algorithm” Based on Correlation between ‘Existing Quality Model, Different Attributes of Defects & Tests’

Usage and nature of software systems have changed significantly. Due to this complexity of software systems has also grown exponentially. In these ever-changing requirements and environment in which software system is being used, maintaining quality of software system is very challenging and difficult. If user requirements are not met as expected, it’s called defect. To improve quality, it’s critical to understand and analyze these defects. In this study root cause analysis technique is used to analyze defects and their attributes, root cause of defects and corrective actions of defects. A quality model is designed based on defects, root cause of defects and tests. A quality algorithm is designed in this study depending on existing quality model, defects, tests and their attributes. This quality algorithm is executed on a software system to validate quality model. The results obtained are analyzed to understand the quality of the software system and how it’s different than existing quality model.

Development of quality improvement matrix: An integrated tools for quality improvement

Continuous quality improvement is essential to ensure the competitiveness of manufacturing industries. Design of Experiments, Statistical Process Control and Failure Mode and Effect Analysis are common tools for improving quality. This study proposes a new tool that could encompass quality improvement steps comprehensively, namely Quality Improvement Matrix. This tool integrates several techniques which are House of Quality, Failure Mode and Effect Analysis (FMEA) and Root Cause Analysis. Quality Improvement Matrix covers identification of defects and its root causes, measuring the magnitude of defects (source of defects), formulating alternatives of solutions for reducing root cause of defects and selecting effective solutions to be implemented. Quality Improvement Matrix is developed by following a similar approach as House of Risk. Thus, it also consists of two matrixes that represent two major stages. The first stage is the identification and prioritizing root causes of defect. Then, the second stage is formulating and selecting solutions for improvements.

An information transfer based novel framework for fault root cause tracing of complex electromechanical systems in the processing industry

Abstract As one of the most important approaches for analyzing the mechanism of fault pervasion, fault root cause tracing is a powerful and useful tool for detecting the fundamental causes of faults so as to prevent any further propagation and amplification. Focused on the problems arising from the lack of systematic and comprehensive integration, an information transfer-based novel data-driven framework for fault root cause tracing of complex electromechanical systems in the processing industry was proposed, taking into consideration the experience and qualitative analysis of conventional fault root cause tracing methods. Firstly, an improved symbolic transfer entropy method was presented to construct a directed-weighted information model for a specific complex electromechanical system based on the information flow. Secondly, considering the feedback mechanisms in the complex electromechanical systems, a method for determining the threshold values of weights was developed to explore the disciplines of fault propagation. Lastly, an iterative method was introduced to identify the fault development process. The fault root cause was traced by analyzing the changes in information transfer between the nodes along with the fault propagation pathway. An actual fault root cause tracing application of a complex electromechanical system is used to verify the effectiveness of the proposed framework. A unique fault root cause is obtained regardless of the choice of the initial variable. Thus, the proposed framework can be flexibly and effectively used in fault root cause tracing for complex electromechanical systems in the processing industry, and formulate the foundation of system vulnerability analysis and condition prediction, as well as other engineering applications.

A Comparative Analysis of Taguchi Methodology and Shainin System DoE in the Optimization of Injection Molding Process Parameters

Two well-known Design of Experiments (DoE) methodologies, such as Taguchi Methods (TM) and Shainin Systems (SS) are compared and analyzed in this study through their implementation in a plastic injection molding unit. Experiments were performed at a perfume bottle cap manufacturing company (made by acrylic material) using TM and SS to find out the root cause of defects and to optimize the process parameters for minimum rejection. Experiments obtained the rejection rate to be 8.57% from 40% (appx.) during trial runs, which is quiet low, representing successful implementation of these DoE methods. The comparison showed that both methodologies gave same set of variables as critical for defect reduction, but with change in their significance order. Also, Taguchi methods require more number of experiments and consume more time compared to the Shainin System. Shainin system is less complicated and is easy to implement, whereas Taguchi methods is statistically more reliable for optimization of process parameters. Finally, experimentations implied that DoE methods are strong and reliable in implementation, as organizations attempt to improve the quality through optimization.

Interaction analysis–based information modeling of complex electromechanical systems in the processing industry

Information modeling for complex electromechanical systems in the processing industry is the foundation for system vulnerability analysis, failure propagation mechanism, and fault root cause tracing driven by data analysis. Focusing on the difficulties in information modeling for complex electromechanical systems, a new approach based on interaction analysis and a general framework for information modeling of complex electromechanical systems are proposed. First, the basic structures of the information model are defined based on monitoring data. Second, an improved symbolic transfer entropy method with procedures for optimizing the number of symbols, phase space reconstruction, binary encoding, and decimal decoding are proposed to detect the interaction direction and quantify the interaction strength between different monitoring variables. Third, some optional methods are introduced to simplify and modify models. Finally, a directed-weighted information model for a specific complex electromechanical system is constructed based on information flow. An actual information modeling application of a complex electromechanical system is used to demonstrate the proposed method and compare it with existing methods. This new approach can handle general information modeling problems and overcome the drawbacks of existing methods since all the monitoring variables are used to improve integrity of the model, and prior knowledge about the physical structure and key points’ selection are not required. A unique and complete information model is obtained regardless of the choice of the initial variable. Thus, the proposed method can be flexibly and effectively used in information modeling for complex electromechanical systems in the processing industry and formulate the foundation of system vulnerability analysis, failure propagation mechanism, and fault root cause tracing methodology, as well as other engineering applications.

Estimation of Disturbance Propagation Path Using Principal Component Analysis (PCA) and Multivariate Granger Causality (MVGC) Techniques

Process monitoring and fault diagnosis using the multivariate statistical methodologies has been extensively used in the process and product development industries for the last several decades. The fault in one process variable readily affects all the other associated variables, which makes the fault detection process not only more difficult but also time-consuming. In this study, principal component analysis (PCA)-based fault amplification algorithm is developed to detect both the root cause of fault and the fault propagation path in the system. The developed algorithm projects the samples on the residual subspace (RS) to determine the disturbance propagation path. Usually, the RS of the fault data is superimposed with the normal process variations, which should be minimized to amplify the fault magnitude. The RS-containing amplified fault is then converted to the covariance matrix, followed by singular value decomposition (SVD) analysis, which, in turn, generates the fault direction matrix corresponding...

Manufacturing Analytics and Industrial Internet of Things

Over the last two decades, manufacturing across the globe has evolved to be more intel-ligent and data driven. In the age of industrial Internet of Things, a smart production unit can be perceived as a large connected industrial system of materials, parts, machines, tools, inventory, and logistics that can relay data and communicate with each other. While, traditionally, the focus has been on machine health and predictive maintenance, the manufacturing industry has also started focusing on analyzing data from the entire production line. These applications bring a new set of analytics challenges. Unlike tradi-tional data mining analysis, which consists of lean datasets (that is, datasets with few fea-tures), manufacturing has fat datasets. In addition, previous approaches to manufacturing analytics restricted themselves to small time periods of data. The latest advances in big data analytics allows researchers to do a deep dive into years of data. Bosch collects and utilizes all available information about its products to increase its understanding of complex linear and nonlinear relationships between parts, machines, and assembly lines. This helps in use cases such as the discovery of the root cause of internal defects. This article presents a case study and provides detail about challenges and approaches in data extraction, modeling, and visualization.