基于神经网络的换热站二次回水温度预测方法

PREDICTION METHOD OF SECONDARY BACKWATER TEMPERATURE OF HEAT EXCHANGE STATION BASED ON NEURAL NETWORK

换热站是集中供热系统的中枢,针对集中供热系统的时变性、时滞性及非线性的特点,为了准确获得换热站温度控制系统的设定值,根据实测的换热站数据和气象数据,通过相关性分析建立三个预测特征集,分别构建RNN、DNNs和LSTM三种神经网络预测模型实现对二次回水温度的预测。实验结果表明,影响二次回水温度的主要气象因素为室外温度,三种预测模型都能实现二次回水温度的预测,在数据集Set3上采用RNN预测模型具有更高的预测精度,并且节约能源。

The heat exchange station is the center of the central heating system.According to the time-varying,time-delay and nonlinear characteristics of the central heating system,in order to accurately obtain the set value of the temperature control system of the heat exchange station,this paper establishes three prediction feature sets based on the measured data of the heat exchange station and meteorological data through correlation analysis.And three neural network prediction models,RNN,DNNs and LSTM,were constructed respectively to realize the prediction of secondary backwater temperature.The experimental results show that the outdoor temperature is the main meteorological factor affecting the temperature of the secondary backwater.These three models can predict the temperature of the secondary backwater,and the use of RNN prediction model in the Set3 has higher prediction accuracy and energy saving.