灰色—支持向量机的空调负荷预测

Air-conditioning load prediction based on grey theory and support vector machine

对空调负荷进行准确预测不仅对空调优化控制的意义重大,而且也是实现空调经济运行与节能的关键所在。为了提高建筑空调负荷的预测精度,在分析灰色模型和支持向量机建模特点基础上提出了一种空调负荷组合预测算法。该方法综合了灰色建模计算过程简单以及支持向量机自学习和泛化能力强的优点,能够更加有效地利用样本数据的有效信息,提高模型预测精度。首先,通过灰色建模过程弱化了样本数据的随机因素。然后,对灰色模型输出进行归一化处理及数据重构,以作为支持向量机的输入。最后,通过支持向量机模型的预测得到最终预测结果。将本文所提出的方法应用于福州一栋办公建筑的逐时空调负荷预测中,并与灰色模型及支持向量机模型作比较,证明了组合模型的预测值与实际运行值拟合度最高,平均绝对误差比灰色模型和支持向量机模型分别降低了47.84%和17.39%。该组合预测模型具有较高的预测精度和更好的泛化能力,具有较强的可行性和实用性。

Accurate prediction of air-conditioning load is very important not only for the optimal control of centre air-conditioning system, but also for the economical running and energy saving of air-conditioning system. In order to improve the accuracy of the forecasting of building air conditioning load, a hybrid Gm^-SVM prediction model is established based on the grey theory and support vector machine （SVM）. By combining the advantages of low computation demand of grey theory with the self-organization of SVM, the historical air condition load information is extracted effectively, and then the model prediction precision can be improved. Firstly, the random noise in original load series was weakened through grey modeling procedure. Secondly, the resulted load series were normalized and reconstructed into the input data for SVM model. Finally, the final prediction of the air-conditioning load was obtained by using SVM. The proposed Gm^-SVM model, SVM model and GM model are used for the hourly air-conditioning load prediction of an office building in summer months in Fuzhou area. The simulated results showed that the hybrid model prediction values were most in agreement with the operational values compared to GM model and SVM model, and the absolute error had reduced by 47.84%and 17.39%, respectively. And therefore, it was considered that the Gm^-SVM method had good quality in terms of prediction precision and generalization.