Answers are grounded in this paper and cite their sources.
This paper develops an occupancy-aware continuous-dimming framework that combines occupancy clustering, a single-luminaire illuminance contribution matrix, and a proximal policy optimization (PPO)-guided particle swarm optimization (PSO) search scheduler for indoor public-space lighting control. The framework maps occupant positions to dynamic activity regions, extracts candidate luminaires from a ceiling-plane geometric rule, and evaluates dimming plans through a linear contribution model that includes a daylight term. Simulation results indicate that the framework maintains high lower-bound illuminance compliance under unseen occupancy layouts and normalized daylight inputs while reducing whole-room over-lighting. PPO-PSO achieved 98.5% lower-bound compliance and a 0.359 energy ratio, close to fixed-PSO at 98.4% and 0.361. The added scheduling layer increased matched online optimization time from 0.073 +/− 0.017 s for fixed-PSO to 0.161 +/− 0.029 s for PPO-PSO, about 2.2 times the fixed-PSO timing, and required 17.8 min of offline policy training. These results position PPO-guided scheduling as a useful adaptive component for variable occupancy and candidate-set conditions, while the main contribution remains the structured engineering formulation that links occupancy-region modeling, contribution-matrix-based dimming, daylight compensation, and computational trade-off evaluation.
No takes yet. Share an insight, caveat, or question.
Zang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: