Abstract
Reliability engineering plays an important role in the design, manufacture, maintenance, and replacement of industrial products. Over the last few decades, accelerated degradation testing (ADT) has been largely utilized to shorten test durations, reduce the samples needed, and provide sufficient degradation data to ensure the effective reliability assessment of the concerned products. Meanwhile, performance degradation modeling has been recognized as an essential approach to help researchers and producers understand the health conditions of the deteriorating systems. However, the diversity in reliability tests, degradation models, and statistical analysis techniques has increased the difficulty in selecting appropriate reliability assessment methods in specific scenarios. Besides, there are no systematic reviews focused on modeling and analysis of performance degradation data. Therefore, this paper aims to (1) present ADT fundamentals, including the basic theory, ADT methods, accelerated stress variables, type of acceleration models, as well as ADT optimization, (2) comprehensively review current states and future challenges in degradation modeling, (3) discuss the problem of model mis-specification and compare different approaches for parameter estimation, (4) highlight future opportunities and possible directions deserving further research.
Highlights
The evolving industrial societies have been characterized by the fast pace of newly developed products appearing on the market
Inspired by the research on reliability modeling and analysis of accelerated degradation data, we present a comprehensive overview of the accelerated degradation testing (ADT) technology, degradation modeling, and parameter estimation for reliability assessment
Since ADT optimization is an essential research branch of reliability engineering, we move to discuss the optimal design of ADT governed by different degradation models under several optimization objectives and predefined constraints
Summary
The evolving industrial societies have been characterized by the fast pace of newly developed products appearing on the market. Considering the effects of time-varying coefficients and time-scale transformations, Elsayed et al [82] developed an extended linear hazard regression model with time-dependent parameters for reliability analysis under normal working conditions by employing failure data obtained during accelerated conditions Since these relationships are obtained by purely fitting degradation data without any physical explanations, statistical models may be unsuitable for characterizing life-stress relationships outside the range of the concerned data, less being used for reliability modeling [83]. Li and Jiang [116] employed the drift Brownian motion to model the degradation process with competing failures, in which the sample size and inspection time under each stress level are optimized by minimizing the expected variance of the q-th percentile of the lifetime distribution of the products under normal working conditions. Future researchers are supposed to consider how to construct confidence intervals based on online degradation data
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
