基于CEEMDAN+RF+AdaBoost的短期负荷预测

高精度的短期负荷预测不仅是电力系统运行稳定的关键,也是构建智能电网的必要保证。为提高电力系统短期负荷预测精度,提出了一种基于完整集成经验模态分解(CEEMDAN)、随机森林(RF)和AdaBoost的预测方法。针对传统分解方法不能完整分解原始负荷序列的问题,利用CEEMDAN分解方法为各个阶段的IMF分解信号添加特定的白噪声,通过计算余量信号来获得各个模态分量,然后针对前9个模态分量构建RF预测模型,针对残余量构建AdaBoost预测模型,并对结果进行重构预测,得出未来24h的负荷预测数据。最后将CEEMDAN+RF+AdaBoost方法应用于华中地区的短期负荷预测,在同等条件下,与预测模型CEEMDAN+RF、EEMD+RF+AdaBoost、EMD+RF+AdaBoost、RF及AdaBoost进行试验对比,结果表明所构建预测模型的精度优于其他对比模型,具有很好的理论指导意义和实际应用前景。

Short-Term Load Forecasting Based on CEEMDAN+RF+AdaBoost

High precision short-term load forecasting is not only the key to the stability operation of power systems, but also the necessary guarantee for building smart grid. In order to improve the accuracy of short-term load forecasting in power systems, a forecasting method is proposed based on integrated empirical mode decomposition (CEEMDAN), random forest (RF) and AdaBoost. In order to solve the problem that the traditional decomposition method can not completely decompose the original load sequence, the CEEMDAN decomposition method is used to add specific white noise to the IMF decomposition signals in each stage, and the residual signal is calculated to obtain each modal component. Then the RF pre-test model is constructed for the first 9 modal components, and AdaBoost is constructed for the residual value Forecast model. The results of the reconstruction forecast is implemented to obtain the next 24 hours of load forecasting data. Finally, the CEEMDAN + RF + AdaBoost method is applied to the short-term load forecasting in Central China. Under the same conditions, it is compared with the prediction models of CEEMDAN + RF, EEMD + RF + AdaBoost, EMD + RF + AdaBoost, RF and AdaBoost. The results show that the accuracy of the prediction model is better than the other comparison models, which has a good theoretical guidance and practical application prospects.