Anomaly Detection for Chromaticity shift of High Power White LED with Mahalanobis Distance Approach

摘要

High power white LED (HPWLED) with the benefits of high efficiency, small size, lower power consumption and high reliability has been commercially used as a substitute of the traditional white light sources in the application of general lighting systems, monitor backlighting and so on. Nowadays, in the LED's reliability field, many previous researches paid attentions only on the lumen depreciation failure in LED products, ignoring another common failure mode called chromaticity shift. In this paper, we used a data-driven method based on a multivariate distance measure, Mahalanobis distance (MD), to detect the chromaticity shift anomaly of HPWLED after aging test. The result shows that by dealing MD distributions with Weibull statistical model, a chromaticity anomaly alarm indicator can be established to detect anomaly of chromaticity shift before HPWLED failed.