Multi-objective optimized five-channel LED simulation system using melanopic-centric model and the five-receptor integration model

Balancing visual quality with retinal-photoreceptor-related spectral quantities remains an important challenge in integrative lighting design. Based on CIE S 026/E:2018, this study proposes a numerical multi-objective spectral optimization framework for a fixed five-channel red–yellow–green–cyan–blue LED configuration. Twenty-seven daylight spectral power distributions (SPDs) measured during one clear summer day in Beijing are used as optimization targets, whereas the individual LED-channel SPDs, mixed SPDs, candidate peak wavelengths, and PWM duty cycles are modeled or optimized numerically. A bi-level genetic algorithm is used: the outer layer selects one common set of candidate channel peak wavelengths for the complete daylight dataset, and the inner layer determines time-resolved PWM duty-cycle variables at each sampling time. Two models are compared: the melanopic-centric model and the five-receptor integration model, which simultaneously matches all five α-opic daylight efficacy ratios (DERs). Under the imposed CCT, TM-30 Rf, and Rg constraints, the five-receptor integration model maintains the tracking errors of all five α-opic DERs below 5%, whereas the melanopic-centric model produces larger deviations in several cone-opic and rhodopic channels. Compared with the melanopic-centric model, the five-receptor integration model reduces the comprehensive color deviation by 9.40%. These results demonstrate the numerical feasibility of jointly selecting a fixed five-channel spectral configuration and time-resolved PWM variables for daylight α-opic DER matching.