Traditional regression approaches to Accelerated Destructive Degradation test (ADDT) data have modeled the mean curve as being representative. However, maximum likelihood estimates of the mean model are likely to be biased when the data are non-Gaussian or highly skewed. The median model can be an alternative for skewed degradation data. In this work, we introduce a nonlinear Quantile Regression (QR) approach for estimating quantile curves of ADDT data. We propose an iterative QR algorithm that uses the generalized expectation-maximization framework to estimate the parameters of the nonlinear QR ADDT model, based on the asymmetric Laplace distribution to accommodate non-Gaussian and skewed errors. Using the asymptotic properties of the QR parameter estimates, we estimate variance-covariance matrix for the tau th QR parameters using order statistics and bootstrap methods. We propose a new prediction method of the quantile of the failure-time distribution in the normal use condition. Confidence intervals for the quantiles of the failure-time distribution are constructed using the parametric bootstrap method. The proposed model is illustrated using an industrial application and compared with the existing model. Various quantile curve estimates derived using the QR ADDT model provide a more flexible modeling framework than the traditional mean ADDT modeling approach.

Unexpected fault or failure in the power plant have caused high maintenance costs, the loss of energy production, and even safety issues. Developments in sensor technologies and data analytics have aided proper preventive maintenance actions for the system to improve asset availability and reduce repair costs. Nevertheless, effective condition monitoring of a power plant experiences a considerable nuisance from challenging issues such as inherent data characteristics such as high correlations between process variables, irrelevant information from environmental noises, and system complexity. To resolve these problems, this paper proposes an integrated monitoring scheme for performing efficient corrective actions by identifying the variables related to anomalies in combined cycle power plants. The scheme includes a clustering-based linear discriminant analysis to extract key variables for reducing dimensionality to efficiently handle the data, followed by employing the Mahalanobis depth statistics for anomaly detection and causal analysis via contribution scores. The proposed monitoring scheme is applied to condition monitoring data of a combined cycle power plant in South Korea, which include two types of anomalous operations. The reliability and robustness of the proposed condition monitoring scheme are validated by comparing other state-of-the-art methods. The proposed method shows a potential in efficiently detecting anomalies during operation and even early detecting the precursors of anomalies. It is expected to prevent imminent faults or failures by taking proper actions to relevant key process parameters of combined cycle power plant in advance.

Purpose: With the rapid development of wireless communication, the occurrence of anomalies resulting from malicious network attacks and system overload is also increasing rapidly. Consequently, detecting network traffic anomalies in a network system has become crucial for preventing server downtime. In this study, we proposed a method for detecting anomalies in network traffic data through signal processing and statistical tests.BRMethods: Based on self-similar characteristics of network traffic data, we employed fractional Brownian motion to extract the Hurst exponent as the health index of network traffic data. Additionally, we proposed the index-based change-point monitoring scheme to assess the network’s current status.BRResults: Analysis of actual network traffic data shows that the method based on the Hurst exponent and change-point estimation can effectively detect anomalies early, prior to real traffic outbreak, preventing network traffic failures.BRConclusion: This research introduced a method for assessing abnormalities and detecting change-point in network overload based on the statistical property of long-range dependency, facilitating early detection of network issues.

Automated anomaly detection is the prerequisite to minimize human errors and costs caused by manual inspection. Recently, image-based anomaly detections have gained more attention by widely adopting machine vision systems and computer-aided detections. We propose a classification method using spectral features based on 2D discrete wavelet packet transform under the hierarchical structure of wavelet energies. By capturing the self-similar and long-range dependent characteristics of 2D fractional Brownian field (fBf), wavelet packet spectra are derived to construct a linear model representing the relationship between wavelet energies and resolution levels. 2D DWPT-based energy features effectively preserve irregular oscillations in original images at high-frequency domains as well as at low-frequency domains under a pyramidal structure. In comparison with the existing 2D discrete wavelet transform method, the proposed method shows a potential in efficiently classifying normal and abnormal image data in a numerical example and a real industrial application.

As an advent of smart factory technology, data-driven condition-based maintenance (CBM) is developed to automatically control the production process in engineering field. CBM usually focuses on diagnosing the production status based on real-time data from the sensors. In general manufacturing field, the performance of production equipment gradually decreases due to the wear or deterioration of equipment. To determine if the process is in-control, degradation modeling of observed data from the equipment and its statistical inference is conducted. In this paper, we propose image-based degradation modeling and change-point detection using spatio-temporal process (STP). To describe the deteriorating patterns of image observation, degradation based on spatial and temporal relationship is conducted. At the same time, change-point is estimated to distinguish the degradation under normal and abnormal production status. Through the application to the image stream in real industry, the proposed monitoring scheme `vely conduct the bi-phase representation providing the change-point of manufacturing processes.

Accelerated degradation tests (ADTs) are commonly used to assess the durability of membrane electrode assembly (MEA) components consisting of polymer electrolyte membrane fuel cells (PEMFC) under harsh stress conditions, estimating their lifetime in actual use condition and uncovering their vital degradation mechanisms. ADTs apply mechanically, chemically, or thermally combined stressors to efficiently investigate the durability of MEAs. However, combined stressors for ADTs might cause biased lifetime prediction because major deterioration mechanisms of MEA components are mixed with each other. This work proposes a method to accurately predict the lifetime of MEA through empirical modeling of its performance degradation through ADTs under potential cycle (carbon corrosion) and load cycle tests (electrocatalysts). To simulate operation modes of fuel cell electric vehicles, MEAs are tested under continuous on-off cycle testing (24 h operating and 1 h break) for 5000 h. Degradation patterns of MEAs are first modeled by the empirical model. The relationship between ADTs (potential and load cycle) and continuous on-off condition is then closely examined to accurately predict MEA lifetime under actual operation environments. The proposed idea has a potential to resolve critical durability issues of MEAs by identifying intermingling effects from other constituents. (c) 2022 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Purpose: Because transformers are the primary component for the delivery of electric power, failures but also small irregularities can lead to expensive damage. Wear-out failure, in particular, can be minimized using lifetime analysis. This study focuses on the lifetime estimation of transformers and considers multiple failure-modes. In this way, the reliability of aging equipment can be analyzed individually.BRMethods: We propose a change-point model for the Weibull distribution for multiple failure-modes. By defining the function for a generalized likelihood ratio, the change-point for wear-out failure as well as the parameters for multiple lifetime-distributions can be estimated.BRResults: The proposed approach is applied to real-life data of electric transformers. As a result of the lifetime analysis, the lifetime distribution, including multiple failure-modes and their cutoff points (in time) can be determined.BRConclusion: Using the investigated change-point analysis, both the lifetime and maintenance schedule of transformers with respect to multiple failure-modes can be estimated, and the chance for expensive equipment failure is reduced significantly.

As a part of the smart factory industries, condition-based maintenance (CBM) is developed for monitoring the status of production systems. While existing maintenance technologies predict a certain period of replacement or repair time, CBM diagnoses the real-time status of objects according to the measurement obtained from sensors. Due to the property of manufacturing facilities that their performance degrades over operating time, the system should be maintained as in-control status. In this paper, we propose CBM methodology based on stochastic partial differential equation (SPDE) for production equipment by monitoring the status of products. By modeling the time-series photography of products into SPDE, the estimated results describe the degrading patterns of images in terms of space and time. The application of images in real industry shows that the proposed approach can be effectively conducted for monitoring and detecting the abnormal status.

An accelerated degradation test (ADT) has become a popular method to accelerate degradation mechanisms by stressing products beyond their normal use conditions. The components of an automobile are degraded over time or cycle due to their constant exposure to friction or wear. Sometimes, the performance degradation can be measured only by destructive inspection such as operating torques of return-springs in a bi-functional DC motor system. Plastic deformation of the return-spring causes the degradation of actuating forces for shield movement, resulting in deterioration of the shield moving speed in a headlight system. We suggest a step-by-step procedure for a reliability analysis for a bi-functional DC motor in a headlight system, based mainly on accelerated destructive degradation test (ADDT) data. We also propose nonlinear degradation models to describe the ADDT data of the return-springs. Exposure effects of high temperatures on the return-springs are quantitatively modeled through the ADDT models. We compare the estimation results from both the closed-form expression and Monte Carlo simulation to predict the failure–time distribution at normal use conditions, showing that the lifetime estimation results from the closed-form formulation are more conservative.

A condition-based maintenance (CBM) has been widely employed to reduce maintenance cost by predicting the health status of many complex systems in prognostics and health management (PHM) framework. Recently, multivariate control charts used in statistical process control (SPC) have been actively introduced as monitoring technology. In this paper, we propose a condition monitoring scheme to monitor the health status of the system of interest. In our condition monitoring scheme, we first define reference data set using one-class support vector machine (OC-SVM) to construct the control limit of multivariate control charts in phase I. Then, parametric control chart or non-parametric control chart is selected according to the results from multivariate normality tests. The proposed condition monitoring scheme is applied to sensor data of two anemometers to evaluate the performance of fault detection power.

Wavelet thresholding (or shrinkage) attempts to remove the noises existing in the signals while preserving inherent pattern characteristics in the reconstruction of true signals. For data-denoising purpose, we present a new wavelet thresholding procedure which employs the step-down testing idea of identifying active contrasts in unreplicated fractional factorial experiments. The proposed method employs bootstrapping methods to a step-down test for thresholding wavelet coefficients. By introducing the concept of a false discovery error rate in testing wavelet coefficients, we shrink the wavelet coefficients with p-values higher than the error rate. The error rate controls the expected proportion of wrongly accepted coefficients among chosen wavelet coefficients. Bootstrap samples are used to approximate the p-value for computational efficiency. We also present some guidelines for selecting the values of hyper-parameters which affect the performance in the step-down thresholding procedure. Based on some common testing signals and an air-conditioner sounds example, the comparison of our proposed procedure with other thresholding methods in the literature is performed. The analytical results show that the proposed procedure has a potential in data-denoising and data-reduction in a variety of signal reconstruction applications.

Since machine vision systems (MVS) lead to a wide usage of monitoring systems for industrial applications, the research on the statistical process control (SPC) of image data has been promoted as an automated method for early detection and prevention of unusual conditions in manufacturing processes. In this paper, we propose a non-parametric SPC approach based on the 2D wavelet spectrum (WS-SPC) to extract the feature that contains the spatial and directional information of each subspace in an image. Using the 2D discrete wavelet transform and spectrum analysis, the representative statistic, the Hurst index, is calculated, and a single matrix space that consists of estimated statistics is reconstructed into a spatial control area for SPC. When a control limit is determined by the density of statistics, real-time monitoring based on WS-SPC is available for time releasing images. In the application, an analysis of wafer bin maps (WBMs) is conducted at a semiconductor company in Korea in order to evaluate the performance of the suggested approach. The results show that the proposed method is effective in terms of its fast computation speed and spectral monitoring.

* 이 논문은 2018년도 정부(교육부)의 재원으로 한국연구재단의 지원을 받아 수행된 기초연구사업임(No. 2018R1D1A1A09083149). 본 연구는 한국전력공사의 2017년 선정 기초연구개발 과제 연구비에 의해 지원되었음(과제번호: R18XA06-46).