A Neural-Network-Based Color Control Method for Multi-Color LED Systems

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journalpeer-review

19 Scopus Citations
View graph of relations



Original languageEnglish
Article number8531652
Pages (from-to)7900-7913
Journal / PublicationIEEE Transactions on Power Electronics
Issue number8
Online published12 Nov 2018
Publication statusPublished - Aug 2019


A neural-network-based color control method for multi-color LED systems is proposed. The proposed control method can achieve color control with high color rendering indexes even when there is ambient light. The spectral peak control provides data points with high color rendering indexes. These data points would be updated and utilized for the training of neural networks on MCU. The trained neural networks can achieve the color control of multi-color LED systems with or without ambient light. Compared with previous color control methods, the proposed method 1) achieves color control with high color rendering index with or without ambient light, 2) does not require an accurate electrical-optical system model, 3) is applicable to all multi-color LED systems, and 4) improves the performance without modifying any hardware. The proposed color control method is tested on an application example of a red-green-blue-amber (RGBA) LED system. The RGBA LED channels are connected in series and driven by a buck-type single-inductor-multiple-output (SIMO) LED driver. The effectiveness and robustness of the proposed control method under different types of ambient lights have been verified experimentally.

Research Area(s)

  • Ambient light, lighting control, multi-color light-emitting diode (LED) system, red-green-blue-amber (RGBA) LED system, single-inductor-multiple-output (SIMO) LED driver