Design of Wide Neural Network Model for Controlling Tunable LED Luminaire: (Neural Network based Luminaire Control)

This paper explores machine learning models to design adjustable light spectra for supporting circadian entrainment using LED luminaires. The approach combines practical experimental work with sophisticated machine learning methods to enhance lighting solutions. There is a growing demand for adjustable sources to improve efficiency and reduce costs, leading to the development of multichannel LED drivers capable of operating at different brightness levels and spectra for each LED channel. The multichannel LED driving circuitry is a critical component of human-centric lighting, which aims to enhance the well-being and health of occupants. The experiment utilized a four-channel warm white and cool white combination LED luminaire under dimming and Correlated Colour Temperature (CCT) variation. Current-controllable LEDs for different CCTs and intensities were tested with the help of DMX 512 Decoder and master control. The Wide Neural Network model designed outperforms other models in performance, helping to determine the current requirement of the lighting control system to set the desired circadian entrainment conditions.