论文部分内容阅读
提出了一种基于小波分解-支持向量机(WD-SVM)的办公建筑空调负荷预测建模方法,利用小波分解将具有较强随机性和非线性的空调负荷信号进行分解,然后利用支持向量机对分解后不同频率下的分支数据进行预测建模,从很大程度上避免了由于训练样本不完备而导致的支持向量机预测精度波动。仿真结果表明,WD-SVM方法的预测精度评价指标EEP比单支持向量机法降低33.6%,预测精度有明显提升。
This paper proposes a method of building air conditioning load forecasting based on Wavelet Decomposition and Support Vector Machine (WD-SVM), decomposes the air conditioning load signal with strong randomness and nonlinearity by wavelet decomposition, and then uses support vector machine Predicting the branching data under different frequencies after the decomposition can largely avoid the fluctuation of the prediction accuracy of the SVM due to the incomplete training samples. The simulation results show that the predictive accuracy evaluation index EEP of WD-SVM method is reduced by 33.6% compared with single support vector machine method, and the prediction accuracy is obviously improved.