基于RBF神经网络的商场建筑空调能耗预测

Prediction of energy consumption of shopping mall air-conditioning based on RBF neural network

商场建筑夏季空调能耗占总能耗的50%以上,鉴于空调能耗较高,对空调能耗进行预测有利于提升运行经济性。针对商场建筑空调系统非线性、多变量等问题,提出一种基于RBF神经网络空调系统能耗预测模型。该方法将日最高温度、日最低温度、日平均温度、日最高湿度、日最低湿度、日平均风速和空调能耗作为RBF神经网络的输入,建立空调系统能耗预测模型,并通过测试数据对精度进行验证。实例表明:预测值和实际值的相对误差为5.96%,均方根误差为1642.7kWh,预测精度高,稳定性好,可满足商场建筑中央空调能耗预测的实际应用要求。

In summer,the energy consumption of air-condition system in shopping malls accounts for more than 50%of the total energy consumption.The high energy consumption of air-condition system is becoming increasingly prominent.Predicting the energy consumption of air-condition is conducive to the energy-saving operation of air conditioning systems.Focus on the non-linearity and multi-variables in the air-condition system of shopping malls,this paper proposes an air-condition system energy consumption prediction model based on RBF neural network.This method puts the maximum temperature,minimum temperature,average temperature,maximum humidity,minimum humidity,wind speed and air conditioning energy consumption into the RBF neural network to establishes an air-condition system energy consumption prediction model,and verifies the accuracy through test data.The proposed example shows that the relative error between the predicted value and the measured value is 5.96%,with a root-mean-square error of 1642.7 kWh,the high precision and high stability of this model can meet the practical application requirements of energy consumption prediction for the central air air-condition of shopping malls.