An Archimedean Copula Function-Based Prediction Method for High-Power White LED Considering Multi-Performance

Abstract

Lumen, chromaticity, and correlated color temperature CCT are the three main performance indicators of failure of high-power white light-emitting diodes (LEDs) (HPWLEDs). The existing data-driven (DD) method to predict the life of LEDs often considers that three performance indicators are independent, or only one or two performance indicators are used for prediction, which may not be accurate enough. To solve this problem, a method based on Archimedean Copula function is proposed to describe the correlation of multiple performance indicators. The test data reported from LUXEON is used as an example to establish a 3-D life prediction model by combining the thresholds of three performances to predict the life of LEDs, and the application of Copula function is explained. The results show that there is a little influence on life prediction of LEDs among the different Copula function structures, while the difference between considering single performance and multi-performance is obvious. The mean time to failure (MTTF) of which is proven to be more conservative than the L70 life predicted by the conventional method.