In-Situ early anomaly detection and remaining useful lifetime prediction for high-power white LEDs with distance and entropy-based long short-term memory recurrent neural networks

High-power white light-emitting diodes (LEDs) have demonstrated superior efficiency and reliability compared to traditional white light sources. However, ensuring maximum performance for a prolonged lifetime use presents a significant challenge for manufacturers and end users, especially in safety–critical applications. Thus, identifying...