基于EMD和SVM的电网工程设备价格预测模型

电网工程设备价格的非线性和非平稳性特征导致其价格预测难度大、预测精度低,针对这一问题,建立了EMD-SVM预测模型。利用经验模态分解（EMD）将历史价格分解为平稳的、周期波动的若干价格分量,并以此作为输入,对各分量进行基于支持向量机（SVM）的价格预测,最后将各预测分量叠加得到预测值。以180 MVA主变的历史数据为样本,通过与SVM的预测结果进行对比及误差分析,验证了EMD-SVM预测方法能够有效提高电网工程设备价格的预测精度,对于工程造价管控和设备招投标具有一定的参考价值。

EMD-SVM-Based Price Forecast Model for Grid Equipment

The equipment price of a grid project is nonlinear and non-stationary, and it is difficult to be estimated accurately. Therefore, the EMD-SVM model is built to divide the historical prices into smooth and periodically-fluctuating components with empirical mode decomposition(EMD). Take these components as input, their forecast values are obtained with support vector machine(SVM), and then these values are superimposed to gain the final forecast price. The comparison between the forecast price and the historical data of a 180-MVA transformer and the error analysis are carried out, which indicates that the EMD-SVM-based forecastion can effectively improve the accuracy of the equipment price forecast and is helpful for project cost control and equipment bidding.