Machine Learning Driven Correlated Color Temperature Tunable LED Lighting System for Indoor lighting applications

This article presents the development of a Correlated Color Temperature (CCT) tunable light system, as most existing systems are more complex and require 4 to 6 or even more different types of Light Emitting Diode (LED) lights for CCT tuning. The objective of this implementation is to optimize indoor lighting for user comfort and energy efficiency, a machine learning (ML) driven system has bee developed for tunable LED lights with Correlated Color Temperature that dynamically adjusts the level of artificial light and CCT based on the ambient daylight or user preferences. A color sensor has been used to measure the CCT of daylight, allowing for the simulation of natural daylight behavior. The CCT variation was produced by mixing the emissions of warm white and cool white LEDs. The intensity of each type of LED was individually controlled using pulse-width modulation (PWM) signals to vary their respective intensities. In turn, a ML technique linear regression (LR) algorithm is used to train all the collected dataset to produce the desired CCT. Additionally, the CCT transitions that mimic natural daylight variations from 6 AM to 6 PM in accordance with the collected dataset. The prototype has been tested with the proposed model which provides high precision with RMSE, R-square and MSE values for warm white LED and cool white LED. The overall system has the ability to replicate typical daylight CCT profiles considerably improves visual comfort, productivity for well being in an indoor environment.