Perspective estimation of light emitting diode reliability measures based on multiply accelerated long run stress testing backed up by stochastic diffusion process

•Multiple stressing loads used in accelerated experiments with very long run times.•Significance of time compression in accelerated tests expanded and explained.•Critical threshold for LED observed characteristics found to reflect current stress level.•First passage time of LED degradation described in the form of a distribution.•Rich set of LED reliability measures and first passage time numerical characteristics obtained. Light emitting diodes (LEDs) are currently a key element of many systems. They provide main, informational, safety, support and other functions. Their design, size and low-cost production make them increasingly popular. The relatively long technical life of electronic components is not always easy to prove and properly estimate, and the correlations between luminance, luminous flux intensity and other electrical characteristics of an LED (e.g. voltage) are not usually available. To ensure that their operation is correct and safe, LEDs must have a suitable level of reliability; LEDs are examined in service to detect their degradation and the deterioration of their technical condition. This article is devoted to estimating and demonstrating some LED reliability measures, such as the time during which the LED operation is failure-free and at the required luminance level. Existing research is typically based on short testing periods, one stress factor and more traditional models for analysing recorded data. In contrast, our approach is based on multiply accelerated reliability testing with very long experiment runs supported by degradation modelling using a stochastic diffusion process. Multiple acceleration is provided by intensifying environmental climatic effects (heat exposure) and simultaneously increasing the stress of operating conditions (increased current load). The parameters for degradation models based on stochastic diffusion processes are estimated for the studied LEDs by a specific form of the maximum likelihood estimation (MLE) method, including parameter confidence intervals. Based on the results, a novel correlation between luminance level, luminous flux intensity and voltage magnitude is also determined. The aim is to estimate and demonstrate key LED reliability measures.