摘要
为了达到减少建筑能耗的目的,提出一种基于神经网络的建筑节能预测方法,以较小的代价精确预测建筑的热环境变化规律。研究结果表明,基于前向神经网络建立的建筑节能预测方法只需要安装15个温度传感器便能精确预测101个房间的建筑温度和热功率,大幅度地节省了成本,其预估的误差在6%左右,准确度达到90%以上,具有潜在的应用价值。后期可以将该技术应用于建筑智能管理系统,向居住者反馈建筑热环境信息,培养居住者生态消费习惯,为可持续发展提供支持。
In order to reduce energy consumption of building, a forecasting method of energy-saving building based on artificial neural network is proposed to predict the variation of building＇s thermal environment with a small cost. The result shows that the forecasting method based on back propagation artificial neutral network is able to predict the temperature and the thermal power of 101 rooms, with only the installation of 15 temperature sensors. It significantly decreased the cost, besides, the estimated error is about 6% and the accuracy is above 90%, which means a value of potential application. In later period, this technique is available for intelligent building management systems, providing the feedback information of thermal environment to occupants, developing their eco-consumption habit and supporting sustainable development.
作者
高扬
陈坦
胡海涛
Gao Yang Chen Tan Hu Haitao(Refrigeration and Cryogenics Institute of Shanghai Jiao Tong University, Shanghai, 200240 Robotics Institute of Shanghai Jiao Tong University, Shanghai, 200240)
出处
《制冷与空调（四川）》
2017年第1期9-13,63,共6页
Refrigeration and Air Conditioning
关键词
神经网络
热仿真模型
节能预测
热环境变化
建筑智能管理系统
Artificial Neural Network
Model of thermal simulation
NEnergy-saving forecast
Variation of thermal environment
Intelligent building management systems
