基于GM-BP神经网络的校园建筑能耗预测

针对季节更迭、教学活动等因素对校园公共建筑能耗的影响,通过建立GM-BP神经网络组合预测模型,借助MATLAB软件完成建模和仿真环节,对建筑能耗开展预测分析研究。同时,引入最大相对误差绝对值Emax、平均相对误差Eave和均方根误差RMSE 3个性能指标对各预测模型性能进行评价。结果表明,组合模型较单一的GM(1,1)模型和BP神经网络模型预测精度更高,拟合性能更好。研究成果对能源管理部门制定用能政策及科研院校从事建筑节能研究具有一定的借鉴意义。

Energy Consumption Prediction of Campus Building Based on the GM-BP Neural Network

A GM-BP neural network combination forecasting model is designed to eliminate the effect of the seasonal alternation, teaching activities and other factors on public building energy consumption on campus. MATLAB software is utilized to complete the modeling and simulation of building energy consump-tionto carry out the forecast analysis. At the same time, to evaluate the performance of the forecast model, the absolute value of maximum relative error(Emax), the average relative error(Eave) and root mean square error(RMSE) are adopted. The results show that the combined model is better than the single GM(1, 1) model and BP neural network model, and the fitting performance is better. The research results have some reference value for the energy management department to formulate energy policy and the research institutions to research on building energy saving.