基于小波分解与随机森林的短期负荷预测

Short Term Load Forecasting Based on Wavelet Decomposition and Random Forest

为解决训练样本少、负荷波动较大,采用数据挖掘方法(神经网络、支持向量机以及随机森林)预测负荷精度不高的问题,文章提出了一种基于小波分解与随机森林结合的负荷预测方法。首先采用小波分解算法将历史负荷序列分解成若干个不同频率的子序列,结合实时气象数据,根据各个频段的负荷子序列的特征,利用随机森林回归算法分别对负荷子序列进行建模,最后将待预测日不同频率分量进行重构得到预测结果。实例中的数据来源于安徽某地的历史负荷,将所提方法与神经网络、支持向量机和随机森林等数据挖掘方法进行比较,证实了所提方法的有效性。

In order to solve the problem of low accuracy of load forecasting using data mining methods (neural network, support vector machine and random forest) with fewer training samples and larger load fluctuation, a load forecasting method based on combination of wavelet decomposition and random forests(IRF) is proposed in this paper. Firstly, the historical load sequence is divided into several subsequences of different frequency bands by using the wavelet decomposition algorithm. Then the load sub-sequences are modeled respectively by using the random forest regression algorithm combined with real-time meteorological data. Finally, the different predicted frequency components are reconstructed. Compared with the traditional BP neural network, support vector machine and random forest, the proposed method has higher forecasting accuracy.