GM-LSSVM模型在建筑能耗预测中的应用
Forecasting of building energy consumption based on grey theory and least squares support vector machine

为提高大型公共建筑能耗的预测精度，提出一种基于灰色模型和最小二乘向量机方法(GM-LSSVM)的办公能耗预测模型. 该方法结合灰色建模计算简单的特点，以及最小二乘支持向量机非线性拟合能力和泛化能力强的优势，充分发掘样本数据的规律，并以粒子群优化算法进行模型参数选择. 根据福州某大型公共建筑能耗数据，通过本研究提出的方法建立预测模型，并与神经网络模型以及最小二乘支持向量机模型的预测结果进行比较，验证了该方法具备较高的预测精度和较强的泛化能力.
In order to improve the predictive accuracy of the building energy consumption model，a hybrid of GM-LSSVM prediction model is established. This method combines the advantage of low computation demand of grey theory and the ability of nonlinear mapping of least squares support vector machine(LS-SVM)，the historical building energy consumption information is extracted effectively，and particle swarm optimization(PSO) is used to select parameters of LS-SVM model. According to the energy consumption data of the public building in southern city，the GM-LSSVM is used to predict the building energy consumption. The results show that the proposed model has higher accuracy and stronger generalization ability than RBF model and LSSVM model