城市居住建筑能耗预测模型

Energy Consumption Predicting Model of Urban Residential Building

针对城市居住建筑能耗问题,利用相关分析、鱼骨图模型得出城市能耗与影响因素联系程度,并结合SVM原理构建城市能耗预测模型。鱼骨图以城市能耗为原因问题,以城市能耗直接相关的因素作为原因,间接相关作为子原因,相关系数作为联系值,得出相关的关系系数并作为因素权重。以28组工程项目数据作为训练样本,利用SVM模型建立相应的预测模型,并采用平均绝对误差进行评估,再以5组测试样本进行检验。研究结果表明,夏温度设定与各因素的相关性极弱;户主年龄较重要,电脑台数较不重要。

Due to energy consumption problem of urban residential building, based on correlation analysis and fishbone diagram, the relationship among energy consumption and indexes was gained and the predicting model was constructed by SVM. About fishbone, energy consumption was set the reason problem, the direct or indirect correlations with the reason problem were set reasons or sub - reasons separately. SVM model were obtained through training of 28 sets of measured data, the mean absolute error was used to verify the stability of the model, and the model was used to discriminate 5 new samples. The results showed that summer temperature setting was very weak correlation with various factors, age was more important, the number of the household computer was less important.