基于AP-BP神经网络的建筑能耗分析与预测

建筑行业对能源的节约是当前节约能耗的重要途径之一，在对能源浪费现状和建筑物能源绩效调研和分析的基础上，提出了一种基于affinity propagation（AP）聚类的back propagation（BP）神经网络建筑能耗分析与预测方法。通过AP聚类算法对影响建筑能耗的多维因素进行聚类分析，得到影响建筑能耗的主要因素并作为BP神经网络的输入，然后将建筑能耗指标热负荷和冷负荷作为BP神经网络的输出，建立建筑能耗分析与预测模型。均方根误差（RMSE）和平均相对泛化误差（ARGE）评价指标分析结果表明，本文所提方法对能耗值预测的拟合程度优于经典的BP神经网络，且通过建筑能耗输入输出的结构调整能够节约能耗，提高能效。

Analysis and prediction of the energy consumption of buildings based on a back propagation-affinity propagation (AP-BP) neural network

Energy saving is one of the most important ways to reduce the energy consumption of buildings. Based on an analysis and investigation of energy waste and the energy efficiency of buildings, a back propagation (BP) neural network based on the affinity propagation (AP) clustering algorithm (AP-BP) for analysis and prediction of the energy consumption of buildings is proposed. By means of the AP clustering algorithm, the key factors affecting the energy consumption of buildings can be obtained, and these are used as inputs of the BP neural network. The thermal load and the cold load of buildings are used as outputs of the BP neural network. Finally, a model for analyzing and predicting the energy consumption of the building is obtained. By evaluating the root mean square error (RMSE) and the mean relative generalization error (ARGE), the experimental results show that the proposed method is better than the classical BP neural network in terms of the degree of fitting of the energy consumption prediction. Using the AP-BP can save energy and improve energy efficiency by adjusting the inputs and outputs of buildings.