A physical–statistical model of overload retardation for crack propagation and application in reliability estimation

Estimation on Reliability Models of Bearing Failure Data

“…If the predicted reliability value is too low, the product function cannot be fully exploited, making the product lose its environmental adaption and leading to huge waste of conditional resources. Therefore, in order to guarantee the safe and stable operation of product system, it is essential to implement effective monitoring and diagnosing and precise reliability model estimation for bearings [3][4][5][6][7].…”

Applications of outlier and change detection for longitudinal data

“…Following this idea, in Chapter 3 we further propose to extend the step changes in the models of Tsiamyrtzis and Hawkins (2005) and Apley (2012) to a piecewise linear model that can be applied to general degradation signals. In many degradation models, a failure limit is defined so that a failure occurs if the degradation path crosses the failure limit (Lu and Meeker, 1993;Gebraeel et al, 2005;Chen and Tsui, 2013;Si et al, 2016). If the control limit of our model is set as the failure limit of a degradation signal, the proposed procedure can be applied to detect failures based on a noisy degradation signal.…”

“…The latter one is also called overloads. Si et al (2016) studied the problem of overload retardation, which occurs in some materials where overloads decelerate rather than accelerate the crack propagation process. The analysis of the overload retardation effect is considered critical since ignoring such effect can lead to a biased estimation of product life.…”

“…The analysis of the overload retardation effect is considered critical since ignoring such effect can lead to a biased estimation of product life. which is based on the experiment conducted in Si et al (2016). In the experiment, the Compact Tension (CT) fatigue loads (Sih, 1973) mixed with a single load are added at cycle 0 on some testing specimens to obtain their crack propagation paths.…”

“…In the experiment, the Compact Tension (CT) fatigue loads (Sih, 1973) mixed with a single load are added at cycle 0 on some testing specimens to obtain their crack propagation paths. Given the observed crack propagation paths, Si et al (2016) proposed a physical-statistical model that is based on a modified Paris law model and a maximum likelihood estimation (MLE) approach to obtain the fitted crack propagation paths with no random errors, which will be further used to determine whether the testing material has an overload retardation effect. The crack propagation path in Figure 3-3(a) is one of the fitted crack propagation paths with no random noise that is obtained in Si et al (2016).…”