Abstract
In recent years, with the improvement of people’s living standards, lamps have played an irreplaceable role as an essential part of daily life. As an important indicator, the lifetime of lamps directly affects their reliability and performance stability. In order to quickly and reliably determine the conventional lifetime of light-emitting diode (LED) lamps, four sets of constant-stress accelerated degradation tests have been performed on LED lamps. The luminance degradation test data were fitted using the three-parameter Weibull right approximation method (TPWRAM), and each sample’s pseudo-failure time was calculated by combining with the failure criteria. Ultimately, the life prediction of LED lamps was achieved by using the least square method (LSM) and the probabilistic life prediction model under Weibull distribution. The results indicate that the overall average mean absolute percentage error (MAPE) of the fitting curve of luminance degradation data for LED lamps by TPWRAM is only 5.024%, the determination coefficient of the fitting line from the pseudo-failure time based on Weibull distribution and LSM is close to 1, as well as the adjusted determination coefficient, and the residual standard deviation is close to 0, proving that the probabilistic model established by Weibull distribution has high accuracy in predicting conventional life. The luminance degradation experiment and the life prediction model of LED lamps can provide important guidance for life prediction of other photoelectric devices.
