Accelerated Degradation Testing Data Research Articles

Uncertain process-based data integration and residual lifetime evaluation of PCB in airborne equipment with ADT and field data

Accurately evaluating the lifespan of the Printed Circuit Board (PCB) in airborne equipment is an essential issue for aircraft design and operation in the marine atmospheric environment. This paper presents a novel evaluation method by fusing Accelerated Degradation Testing (ADT) data, degradation data, and life data of small samples based on the uncertainty degradation process. An uncertain life model of PCB in airborne equipment is constructed by employing the uncertain distribution that considers the accelerated factor of multiple environmental conditions such as temperature, humidity, and salinity. In addition, a degradation process model of PCB in airborne equipment is constructed by employing the uncertain process of fusing ADT data and field data, in which the performance characteristics of dynamic cumulative change are included. Based on minimizing the pth sample moments, an integrated method for parameter estimation of the PCB in airborne equipment is proposed by fusing the multi-source data of life, degradation, and ADT. An engineering case illustrates the effectiveness and advantage of the proposed method.

Accelerated degradation test method and performance prediction with service mileage of the yaw damper in railway vehicles

The performance of the yaw damper for rail vehicles degrades significantly degradation after long-term operation, which directly affects the dynamic behaviour of the vehicle operation. To quickly achieve the performance degradation of the yaw damper and predict the operating mileage under actual service conditions, the operating displacement spectrum of the yaw damper under actual operating conditions is analysed and decomposed using the short-time Fourier transform. Based on the equivalent principle of energy loss during vibration attenuation, an accelerated test method is proposed for the performance degradation of yaw dampers. By conducting the accelerated degradation test of the yaw damper on the test bench, the continuous degradation relationship between the performance parameters of the damper and the loading times is obtained. By combining the operating displacement spectrum and the accelerated degradation test data, the performance degradation trajectory model of the yaw damper is established, taking into account the equivalent energy loss between the service mileage and the loading times. The mapping relationship among loading times, service mileage and performance degradation is constructed. Finally, parameter performance tests are conducted on yaw dampers under different service mileage to verify the accuracy of the performance degradation trajectory model. The results show that the proposed acceleration test method and degradation trajectory model can quickly realise the performance degradation of the damper and accurately predict the performance parameters for different service mileage under actual operating conditions.

Exploring Initiation-Growth Correlation in Accelerated Degradation Testing Data: Both Classical and Bayesian Approaches

Exploring Initiation-Growth Correlation in Accelerated Degradation Testing Data: Both Classical and Bayesian Approaches

Uncertain random accelerated degradation modelling and statistical analysis with aleatory and epistemic uncertainties from multiple dimensions

Accelerated degradation testing (ADT) is commonly employed for degradation analysis and reliability evaluations. Due to limitations in practical ADTs, aleatory and epistemic uncertainties are simultaneously embodied in time, unit, and stress dimensions but they are not considered comprehensively and integrated scientifically currently. To address such problems, we focus on the most common situation in ADTs that test time is relatively long (reflecting aleatory uncertainties in time dimension) while the numbers of test units and stress levels are quite small (reflecting epistemic uncertainties in unit and stress dimensions), and establish an uncertain random accelerated degradation model (URADM). In the URADM, aleatory and epistemic uncertainties are quantified using probability theory and uncertainty theory, respectively, and they are integrated using chance theory. Then, first passage time and reliability evaluations are derived. Next, a three-step uncertain random statistical method is presented for parameter estimations, where aleatory and epistemic uncertainties are divided carefully and quantified separately. A simulation study and a practical case are conducted to show the effectiveness of the URADM. Results reveal that the URADM can not only predict deterministic degradation trends with high accuracy as existing methods do, but contribute degradation boundaries well covering ADT data with narrower boundaries and higher stability.

Reliability assessment of permanent magnet brake based on accelerated bivariate Wiener degradation process.

Permanent magnet brake (PMB) is a safe and effective braking mechanism used to stop and hold the load in place. Due to its complex structure and high reliability, assessing the reliability of PMB remains a challenge. The main difficulty lies in that there are several performance indicators reflecting the health state of PMB, and they are correlated with each other. In order to assess the reliability of PMB more accurately, a constant stress accelerated degradation test (ADT) is carried out to collect degradation data of two main performance indicators in PMB. An accelerated bivariate Wiener degradation model is proposed to analyse the ADT data. In the proposed model, the relationship between degradation rate and stress levels is described by Arrhenius model, and a common random effect is introduced to describe the unit-to-unit variation and correlation between the two performance indicators. The Markov Chain Monte Carlo (MCMC) algorithm is performed to obtain the point and interval estimates of the model parameters. Finally, the proposed model and method are applied to analyse the accelerated degradation data of PMB, and the results show that the reliability of PMB at the used condition can be quantified quite well.

Coupling simulation and accelerated degradation model for reliability estimation: Application to a voltage regulator

This paper demonstrates the use of numerical simulation data, depending on the behavior of a low voltage dropout regulator, for reliability performance estimation based on the accelerated degradation test data. The regulator was designed using Cadence Virtuoso software in 180 nm AMS CMOS technology, and simulated to evaluate its output voltage variations to both temperature and input voltage. The output voltage degradation data were generated according to environmental parameters (input voltage and temperature) constraints, which makes it possible to define failure thresholds under accelerated conditions, making use of the numerical simulation model along with the proposed degradation model. Degradation path model has been adopted to determine the pseudo failure time under the specified failure criterion (5 %). Acceleration law model has then been derived to estimate the reliability model parameters by performing maximum likelihood estimation method not only to analyse but also to predict lifetime data distribution of the regulator under different voltage and temperature stress conditions.

Lifetime Prediction of Residual Current Circuit Breaker with Overcurrent Protection Based on GA‐BP and GS‐SVM Combined Model

Residual current circuit breaker with overcurrent (RCBO) is an electrical appliance which involves the safety of users. In order to predict its lifetime, a combined prediction method of Genetic Algorithm‐Back Propagation (GA‐BP) and support vector machine of grid search (GS‐SVM) is proposed. First, the constant stress accelerated degradation test (ADT) is carried out for the RCBO, and the ADT data are statistically analyzed to establish the performance degradation trajectory. Then, the GA‐BP model and GS‐SVM model are established to predict the degradation trend of the residual operating current value. Then, based on the GA‐BP prediction model and GS‐SVM prediction model, the optimal weighted combined prediction based on the least square method is established. The combined prediction model is used to predict the pseudo‐failure life under each accelerated stress, and the Arrhenius accelerated model is used to estimate the pseudo‐failure life by the least square method. The service lifetime of RCBO under normal temperature (25 °C) is predicted. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Recoverability effects on reliability assessment for accelerated degradation testing

Accelerated Degradation Testing (ADT) provides an efficient experimental approach to collect lifetime-related data for the reliability assessment of highly reliable products under normal use stress. Recoverability, which occurs in many typical failure modes when online in-chamber measurements have to be replaced by offline ex-chamber measurements with stress released, is an important factor that may affect the accuracy of reliability estimation from ADT data. Nonetheless, the presence of recoverability has not been adequately considered in traditional methods, leading to an over-optimistic estimation of lifetime and reliability. The linkage between recoverability and such inferential bias has not been theoretically studied systematically. In this study, recoverability is explicitly incorporated into the ADT modeling framework. Without loss of generality, the Wiener process is adopted as the basis for the proposed degradation model, superimposed with a cumulative recovery. Theoretical results show that the Mean Time To Failure (MTTF) will be overestimated with recoverability neglected, which leads to poor lifetime-centered decision-making. For finite and even small sample sizes, this conclusion is not intuitively certain but the chance is still high and the corresponding overestimation probability can be calculated explicitly. All theoretical conclusions are validated by simulation studies, in which the MTTFs are overestimated from 6% to 42% under different parameter settings. The real-world application of semiconductor products shows that even slight recoverability could lead to an obvious overestimation of MTTF and the proposed model provides a convenient way to derive more accurate assessment results.

Uncertain differential equation based accelerated degradation modeling

Due to economic and technical reasons, in accelerated degradation testing (ADT) there are usually limited ADT data, which embody lots of epistemic uncertainties that cannot be depicted well by probability theory. Noting these facts, several uncertain processes based accelerated degradation models (UADMs) are proposed under the framework of uncertainty theory. However, these UADMs described uncertainties from a macro perspective which ignored the characteristic of performance degradation in small time scales and failed to describe the essence of nonlinear degradation. What is more, to estimate unknown parameters, these models used empirical distribution functions, which may have a great impact on the estimation results with limited samples. Motivated by these problems, based on uncertainty theory this paper builds up an uncertain differential equation based accelerated degradation model (UDEADM), which considers the dynamic cumulative change process of performance in small space–time scale, and explains the essence of nonlinear for uncertainties in degradation trend. Unknown parameters are estimated using uncertain generalized moment estimations. Furthermore, reliability and lifetime are analyzed under the normal operating condition based on belief reliability theory. A simulation study and a real data analysis are conducted to illustrate the effectiveness and advantages of the proposed methodology in details.

Deep-Learning-Based Remaining Useful Life Prediction Based on a Multi-Scale Dilated Convolution Network

Remaining useful life (RUL) prediction of key components is an important influencing factor in making accurate maintenance decisions for mechanical systems. With the rapid development of deep learning (DL) techniques, the research on RUL prediction based on the data-driven model is increasingly widespread. Compared with the conventional convolution neural networks (CNNs), the multi-scale CNNs can extract different-scale feature information, which exhibits a better performance in the RUL prediction. However, the existing multi-scale CNNs employ multiple convolution kernels with different sizes to construct the network framework. There are two main shortcomings of this approach: (1) the convolution operation based on multiple size convolution kernels requires enormous computation and has a low operational efficiency, which severely restricts its application in practical engineering. (2) The convolutional layer with a large size convolution kernel needs a mass of weight parameters, leading to a dramatic increase in the network training time and making it prone to overfitting in the case of small datasets. To address the above issues, a multi-scale dilated convolution network (MsDCN) is proposed for RUL prediction in this article. The MsDCN adopts a new multi-scale dilation convolution fusion unit (MsDCFU), in which the multi-scale network framework is composed of convolution operations with different dilated factors. This effectively expands the range of receptive field (RF) for the convolution kernel without an additional computational burden. Moreover, the MsDCFU employs the depthwise separable convolution (DSC) to further improve the operational efficiency of the prognostics model. Finally, the proposed method was validated with the accelerated degradation test data of rolling element bearings (REBs). The experimental results demonstrate that the proposed MSDCN has a higher RUL prediction accuracy compared to some typical CNNs and better operational efficiency than the existing multi-scale CNNs based on different convolution kernel sizes.

State of Health Prediction of Lithium-Ion Batteries Using Accelerated Degradation Test Data

Lithium-ion batteries are the core equipment of electric vehicles (EVs) and they are expected to spread out through smart power systems [e.g., smart grids (SGs)]. Batteries are subject to degradation over time, which gradually reduces the supplied voltage and capacity until they fail to meet minimum requirements. Battery prognostic is mandatory in order to keep track of the degradation over time and to schedule maintenance and replacement before failures, which can cause severe damages and harmful effects on EVs and SGs. However, the estimation of the state of health and the remaining useful life of batteries is not trivial, since neither can be directly measured. In this article, multiple linear regression models are developed within deterministic and probabilistic frameworks to predict the SoH of batteries based on historical charge/discharge data. The models exploit informative data collected for the battery under investigation and stored data of batteries characterized by accelerated degradation tests. Numerical experiments based on actual data are presented to validate the performance of the proposal.

Bearing Prognostics: An Instance-Based Learning Approach with Feature Engineering, Data Augmentation, and Similarity Evaluation

We propose an instance-based learning approach with data augmentation and similarity evaluation to estimate the remaining useful life (RUL) of a mechanical component for health management. The publicly available PRONOSTIA datasets, which provide accelerated degradation test data for bearings, are used in our study. The challenges with the datasets include a very limited number of run-to-failure examples, no failure mode information, and a wide range of bearing life spans. Without a large number of training samples, feature engineering is necessary. Principal component analysis is applied to the spectrogram of vibration signals to obtain prognostic feature sequences. A data augmentation strategy is developed to generate synthetic prognostic feature sequences using learning instances. Subsequently, similarities between the test and learning instances can be assessed using a root mean squared (RMS) difference measure. Finally, an ensemble method is developed to aggregate the RUL estimates based on multiple similar prognostic feature sequences. The proposed approach demonstrates comparable performance with published solutions in the literature. It serves as an alternative method for solving the RUL estimation problem.

Optimal design of hybrid accelerated test based on the Inverse Gaussian process model

Accelerated life testing (ALT) and accelerated degradation testing (ADT) are two widely used accelerated testing (AT) methods for reliability evaluation of highly reliable products. However, in many cases, it is difficult to meet the estimation requirements only using ALT or ADT because of the nature of data and the test constraints. Since both ALT and ADT data provide useful reliability information, it would be quite valuable to use both of them for reliability estimation. In this paper, an optimal testing plan is proposed for the first time for such a hybrid AT under several experimental design constraints. An Inverse Gaussian (IG) process and the corresponding lifetime distribution are used to model the degradation process and the lifetime of the product, respectively. The strategy of allocating the test units to ALT and ADT and determining the data collection strategy in the experiment are presented. The objective is to maximize the estimation precision of the p-quantile of the product's lifetime under the normal use condition. A study on the stress relaxation of a type of electrical connector is conducted to illustrate the value of the proposed method in practice.

The Evaluation Method for Step-Down-Stress Accelerated Degradation Testing Based on Inverse Gaussian Process

In view of the disadvantage that traditional evaluation method of accelerated degradation testing (ADT) complex statistical analysis, in this paper, the randomness and monotonicity of the degradation path for products with high reliability and long lifetime are taken into consideration based on Inverse Gaussian (IG) process. The evaluation method of step-down-stress ADT as well as its reliability function are established based on IG process. At same time, both the simple IG process and IG process with random effects are considered, respectively. The maximum likelihood estimation method and Markov Chain Monte Carlo (MCMC) estimation method are presented to estimate the unknown parameters in the proposed degradation models, respectively. Finally, the proposed evaluation methods are demonstrated by the step-down-stress ADT data of a certain type of missile tank. The results show that the proposed evaluation method is reasonable and valid in this paper. In addition, compared with simple IG process model, IG process with random effects model has many superb properties when dealing with random effects of products in ADT.

A doubly accelerated degradation model based on the inverse Gaussian process and its objective Bayesian analysis

ABSTRACT The accelerated degradation test (ADT) is an effective method for evaluating the lifetime of high-reliability products. In this paper, a doubly accelerated degradation model based on the inverse Gaussian process is proposed to characterize the ADT data, and then an objective Bayesian approach is presented to analyze the model. Some important noninformative priors including the Jeffreys prior and reference priors under different group orderings are derived. The propriety of the posterior distributions under each prior is validated. A simulation study is carried out to show the superiority of objective Bayesian approach compared with the parametric Bootstrap method. Finally, the approach is applied to analyze a carbon film data.

Failure Prediction of Hard Disk Drives Based on Adaptive Rao–Blackwellized Particle Filter Error Tracking Method

Active failure prediction of hard disk drives (HDDs) is critical to prevent data loss and spare parts replacement decisions. Existing methods for failure predictions of HDDs always used a binary classifier to distinguish the healthy or failed HDDs and cannot address the problem of variable degradation states. In this article, an adaptive error tracking method is proposed for the HDD failure prediction. This method regards the extracted degradation feature as time serials and uses a state filter to estimate the real-time HDD's health status. Then, the HDD failure online prediction is achieved according to the alarm threshold determined by the adaptive error tracking. The degradation of an HDD is described by a first-order Markov hybrid jump degradation model, and the advanced Rao-Blackwellized particle filter algorithm, together with the expectation-maximization (EM) algorithm, is derived to estimate the model parameters adaptively. Finally, to verify the effectiveness of the proposed method, an accelerated degradation test (ADT) based on the vibration was carried out. And the data from ADT and real data center show that the proposed method performs much better than the previous methods, such as Kalman filter, SVM, MD, and recurrent neural network (RNN) based methods, with respect to failure prediction and the alarm distance, which helps to backup data and optimize maintenance decision costs for users.

Durability of seawater and sea sand concrete and seawater and sea sand concrete–filled fibre-reinforced polymer/stainless steel tubular stub columns

This article presents an experimental investigation on the durability behaviour of seawater sea sand concrete and seawater sea sand concrete–filled fibre-reinforced polymer/stainless steel tubular stub columns. Effects of NaCl of seawater on the strength of seawater sea sand concrete and on the deterioration of fibre-reinforced polymer were studied. Accelerated degradation tests were conducted on fibre-reinforced polymer rings exposed to a combined environment of 3.5% NaCl solution and seawater sea sand concrete. Obvious hoop strength reductions were observed in glass fibre-reinforced polymer and basalt fibre-reinforced polymer rings after 6-month exposure at 60°C. Seawater sea sand concrete–filled glass fibre-reinforced polymer tubular stub columns were exposed to an indoor environment (i.e. aged in air at room temperature) for a maximum duration of 2.5 years and no degradation was found by comparing the axial compressive test results from unexposed and exposed specimens. Seawater sea sand concrete–filled stainless steel tubes did not show any deterioration in strength after a 2.5-year exposure to an indoor environment or a 1.5-year immersion in NaCl solution. This study indicated that a hydrothermal environment (e.g. full immersion in solution) is much more aggressive to fibre-reinforced polymer than a dry environment. The reliability of using accelerated degradation test data to estimate the long-term performance of fibre-reinforced polymer–related structures in a real environment may need further research.

Reliability estimation considering multi-stress monotonic degradation test data with non-constant scale parameter

The accelerated degradation testing (ADT) has been proven to be effective in quickly assessing the reliability of a highly reliable product, especially during product design and development stage. However, the modeling of ADT data to estimate reliability at normal operating conditions is still a challenging task. It becomes even more complex when there are more than one accelerating factors. In this study, a nonstationary gamma process is considered to model the degradation behavior assuming that the nature of product deterioration is strictly monotonic and non-negative. It further assumes that both gamma process parameters are stress-dependent while considering multiple stresses and their interaction effects. To deal with a problem of computational complexity caused by the consideration of multiple stresses and parameter dependency, a maximum likelihood method has been used for the model parameter estimation. The lifetime and reliability parameters under normal operating conditions are estimated using the acceleration models. A Monte Carlo simulation with the Bayesian updating method has been incorporated to update the gamma parameters when new degradation data become available. A carbon-film resistor degradation data is employed to demonstrate the efficacy of the proposed reliability assessment methodology.

Uncertain accelerated degradation modeling and analysis considering epistemic uncertainties in time and unit dimension

Accelerated degradation testing (ADT) has been widely used for the reliability and lifetime evaluations of highly reliable products. Generally speaking, the obtained ADT data are composed of two parts - the deterministic degradation trend and uncertainties. When only limited ADT data can be obtained, there will be a lack of knowledge on recognizing the population and lead to epistemic uncertainties. However, current accelerated degradation models fail to clearly distinguish and properly quantify the epistemic uncertainties in time and unit dimensions. Motivated by this problem, this paper constructs a new uncertain accelerated degradation model based on uncertainty theory and belief reliability theory, and presents the uncertain statistical method for parameter estimations. An application case is used to illustrate the proposed methodology and conduct discussions for the deterministic degradation trend and uncertainty analyses of the proposed methodology and the analyses of the proposed methodology to the data sizes in time and unit dimensions. Results show that under limited data sizes in time and unit dimensions, the proposed model can clearly distinguish and properly quantify the epistemic uncertainties in time and unit dimensions and is more suitable for the ADT modeling and analysis than the other two widely used accelerated degradation models.

A Model-Ranking Approach for Estimation Based on Accelerated Degradation Test Data

Motivated by an accelerated degradation test (ADT) on the power gain of microwave power amplifiers, in this article we propose a model-ranking approach for the estimation of some important reliability characteristics. Different degradation models and statistical lifetime distributions are applied to model the data obtained from the ADT. We study the effect of model misspecification in estimating the reliability characteristics when the behavior of the degradation process and the underlying degradation-data-generating mechanism are unknown. We then propose a model-ranking approach with weighted estimation procedures when multiple candidate models are under consideration. Through a Monte Carlo simulation study, we show that the proposed approach is robust and insensitive to model misspecification. Finally, the ADT data from the motivating example are used to illustrate the proposed methodologies.