中国照明电器

针对传统照明设备在复杂场景下环境感知维度单一、调控策略静态化导致的能耗冗余与体验失衡问题，文章提出计算机视觉驱动的自适应照明设备检测与动态调控方法。通过构建多光谱融合感知网络与时空—语义双流分析框架，突破高反射干扰抑制、设备密集遮挡检测等关键技术；设计约束感知深度强化学习算法，实现能效与舒适度的均衡优化。实验表明，该方法使设备检测平均精度达92.7%，动态场景下系统能耗降低31.5%的同时维持视觉舒适度评分≥8.6。

In response to the issues of redundant energy consumption and experience imbalance caused by the single environmental perception dimension and static control strategies of traditional lighting equipment in complex scenarios, this paper proposes a computer vision-driven adaptive detection and dynamic control method for lighting devices. By constructing a multi-spectral fusion perception network and a spatiotemporal-semantics dual-stream analysis framework, key technologies such as high-reflection interference suppression and dense occlusion detection of devices are overcome; a constrained perception deep reinforcement learning algorithm is designed to achieve balanced optimization of energy e fficiency and comfort. Experiments show that this method achieves an average detection accuracy of 92.7%, reduces system energy consumption by 31.5% in dynamic scenes while maintaining a visual comfort score of ≥8.6.