During the operation of an LED array, its thermal and optical performances are always not equal to the superposition of the individual LED's characteristics because of a significant thermal coupling effect between the arrays. Based on this, this paper proposes an electrical–photo-thermal model, with considering both junction temperature and luminous flux, to predict the both the thermal and optical performances of LED arrays operated under different currents, case temperatures, and lighting methods. The junction temperature and luminous flux of a single LED operating under different driving currents and case temperature conditions are firstly collected to establish the luminous flux response surface model of a single chip. Then it is used to predict the luminous flux of an array, whose junction temperature is predicted using both thermal coupling matrix (TCM) and numerical models. Experiments verify the luminous flux of the LED array under different operation conditions and show that the proposed electrical–photo-thermal modeling can be used to predict the thermal and optical parameters of LED arrays with 95 % accuracy. Thus, it is effective for the fast prediction of the junction temperature and luminous flux of large LED systems with array structures, i.e. intelligent automotive lightings and displays.

@en

Spectral power distribution (SPD) is the radiation power intensity at different wavelengths, containing the most basic photometric and colorimetric performance of the illuminant, which is able to predict the lifetime of LEDs. This paper proposes an SPD model assisted by machine learning algorithms to detect the early failure of white LEDs. The SPD features of 3W high-power white LEDs were firstly extracted by the statistical models of Gaussian, Lorentz, and Asym2sig functions. An unsupervised learning method, principal component analysis (PCA), was then used to reduce the extracted features parameters’ dimensions. Next a K-nearest neighbor (KNN)-based method was used to detect LEDs’ anomalies by dividing the main cluster into groups, and estimating the distance from the center of mass of each cluster to the test point. The results showed the following: (1) for selected white LEDs, the Asym2sig function has a better fitting result than Gaussian and Lorentz functions; (2) machine learning methods can significantly assist in LED anomaly detection and can decrease the amount of anomaly detection time to 789.6 h, compared to the 1311 h when lumen maintenance degradation reaches 70% as required by IES TM21.@en