Bayesian based lifetime prediction for high-power white LEDs

作者：

Highlights：

• Traditional NLS regression-based standards showed higher prediction inaccuracy.

• Accelerated degradation test designed to collect lumen degradation data for LEDs.

• A probabilistic estimation of model parameters with a Bayesian method is presented.

• Applied on two case studies of LED packages and lamps to validate proposed method.

• Proposed method showed an improved accuracy compared with TM-21 and TM-28 methods.

摘要

•Traditional NLS regression-based standards showed higher prediction inaccuracy.•Accelerated degradation test designed to collect lumen degradation data for LEDs.•A probabilistic estimation of model parameters with a Bayesian method is presented.•Applied on two case studies of LED packages and lamps to validate proposed method.•Proposed method showed an improved accuracy compared with TM-21 and TM-28 methods.

论文关键词：Light-emitting diodes (LED),Bayesian methods (BM),Monte Carlo Markov Chain (MCMC),Metropolis Hasting (MH),Lifetime prediction

论文评审过程：Received 28 August 2020, Revised 19 June 2021, Accepted 14 July 2021, Available online 21 July 2021, Version of Record 29 July 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115627