On-board Failure Diagnostics and Failure Prediction of High Power White Light Emitting Diodes Used in Safety Critical Applications

Abstract

Light Emitting Diodes (LEDs) have become a focus of global lighting community owing to their long service life, low energy consumption and environmental friendliness. The ever-increasing use of LEDs in many applications such as indoor lighting, street lighting, displays, signalling etc., has created a greater interest for this novel technology among the business community to invest more. As a result, the research towards boosting the light output and improved efficacy has led to the development of High-Power White Light Emitting Diodes (HPWLEDs) which are considered to be the next-generation green lighting sources. HPWLEDs have the potential to replace the traditional light sources such as incandescent lamps, halogen lamps and fluorescent lamps. The process of diagnosing, monitoring the health and, thereby predicting the reliability and remaining useful life of the systems is the central theme of Prognosis and Health Management (PHM). PHM relies on the continuous monitoring of system’s health parameters to detect the possible indications of any suspicious behaviour from the normal course of performance and thereby predicting the remaining useful service life of the systems. Though LED based lights are known for their long-term service life compared to the traditional tungsten halogen or fluorescent lamp-based lights, in reality, their long term performance relies on the thermal design. The heat generated in the p-n junction would bring down the life of the LED device if not dissipated properly. The end user confidence, especially in a safety critical application like airfield ground lighting, becomes high if the lifetime claims of the light manufacturers are based on real-time usage data. Condition based maintenance schedules would avoid sudden failures and expensive system down time if the operators have access to an early warning system based on Remaining Useful Life (RUL) prediction during the transformation phase to the LED technology and thereafter. This research work discusses the key factors affecting the useful life of high-power LED based lights used in safety critical applications. Model-based prognostic approaches are presented to predict the RUL of the lights that use high power white LEDs and these model-based approaches can be readily implemented on-board the lighting system. The study primarily considers the lumen degradation of the lights over long time as the primary failure criterion. As part of the study, long time lumen maintenance data has been collected through the experiments for various test conditions. Two different methods have been studied to estimate the time to reach the failure threshold and the results obtained from these two methods are analysed. The proposed methods to estimate the RUL of the lights use an algorithm to estimate the life as a function of measurable parameters whose values can be obtained from on-board sensors in a real time environment. The thesis also discusses the failure criterion based on colour shift of LED lights. A detailed analysis has been carried out to understand the basic factors that cause the colour shift in the phosphor converted LED lights. A model has been proposed to study and quantify the colour shifting phenomenon.