A nonlinear degradation path dependent end-of-life estimation framework from noisy observations

For the current advanced technology nodes, the end-of-life and reliability statistics estimation is regarded as a key component of devices dynamic reliability management frameworks. An accurate estimation can enable effective lifetime management via adopting appropriate mission profile specific policies. This paper proposes an end-of-life and reliability estimation framework, which takes into account the nonlinearities of the degradation process, as well as the sensors measurements and degradation process uncertainty, aiming to characterize more realistically the devices aging dynamics. Based on the degradation history, the estimation results are updated adaptively via the Bayesian method, once new degradation measurement data are provided. In order to validate and assess the estimation accuracy of the proposed framework, numerical simulations were performed on a power law degradation model. The obtained results for the considered nonlinear degradation process, reveal that, when compared with commonly employed Wiener processes with linear mean, our approach exhibits improved estimation accuracy. Thus, it may be better suited to capture the nonlinearity and variability of in-field degradation dynamics and further to assess/predict the devices reliability in a more realistic manner. 2013 Elsevier Ltd. All rights reserved.