Toward the Adaptive Control of the Colour Rendering of LED Lighting with an RGBW Sensor

The requirements of lighting installations are becoming more sophisticated and smart lighting systems are now widely available. Such systems are now required to meet specific colour rendering quality indices such as CRI and TM-30. We report here on our use of an inexpensive RGBW sensor to return CRI values to allow automatic adjustment of a lighting system's output to meet a given specification. We achieved this through training of a linear model (canonical correlation analysis) as well as a nonlinear one (feedforward neural network). Results demonstrate that it is not necessary to use an expensive spectrophotometer to measure a lighting system's output to derive advanced perception-driven colour rendering quality indices such as the recent TM-30.