基于SOM映射网络的人群流量预测在节能照明中的应用

通常比较常见的照明控制都是事先设定好的时间控制模式,不可调节且也不能满足精细化控制以及智能化的要求。运用SOM自组织映射网络将人流量历史数据进行特征提取并分类,再将各类数据结果运用BP神经元网络方法进行预测,并将预测结果结合照度需求,不同等级人群流量给予不同等级的照度输出,最后在节能方面也与传统照明方式做了对比。实验结果表明,SOM-BP神经元算法预测下的短期人流量预测数据比BP算法精度更高,结合照明调节后在节能方面具有更好的效果,为照明系统提供了新的节能方案。

Crowd-flow Forecast Based on SOM Neural Network in Application of Energy-saving Lighting

Generally the more common lights -control is set in preset time control mode, which cannot be adjusted and would not meet the requirements of fine and smart control. SOM self-organizing feature map network is used for flow data feature extraction and classification, the results are used as inputs of BP network to make a short-term forecast, and then combined with the intensity of illumination requirements to output the different rank as same level of flow, the comparison with traditional way is made in terms of energy -saving. The experimental results show that flow data forecast based on the SOM combined with BP neural network prediction of short-term is more precise than original BP network, has a better effect in energy saving, and has provided the new energy conservation plan for the lighting system.