基于模态分解及GRU-XGBoost短期电力负荷预测 |
点此下载全文 |
引用本文:冉启武,张宇航.基于模态分解及GRU-XGBoost短期电力负荷预测[J].电网与清洁能源,2024,40(4):18~27 |
摘要点击次数: 92 |
全文下载次数: 41 |
|
基金项目:陕西省自然科学基础研究计划(2023-JC-YB-442) |
|
中文摘要:精确的短期电力负荷预测能有效提高电力系统运营水平。针对电力负荷数据受多种因素影响,波动性和随机性强等问题,提出了一种基于模态分解及混合模型的负荷预测方法。首先,采用主成分分析法(principal component analysis,PCA)对负荷特征向量进行处理,去掉冗余信息,再用完全自适应噪声集合经验模态分解(complete ensemble empirical mode decomposition with adaptive noise,CEEMDAN)将历史负荷分解为简化的几个子序列;其次,选择引入样本熵(sample entropy,SE)来计算子序列熵值,将相近的子序列重构得到随机、细节、低频和趋势分量后选用不同结构门控循环单元(gate recurrent unit,GRU)对不同分量类型进行预测,再使用极致梯度提升模型(extreme gradient boosting,XGBoost)对各分量残差进行拟合,各重组序列的预测值为GRU预测值与XBGoost拟合值之和,重组各序列得到最终预测值。选取3年时电力负荷数据进行实验,结果表明,所提模型的均方根误差(root mean square error,RMSE)、平均绝对百分比误差(mean absolutepercentage error,MAPE)和平均绝对误差(mean absolute error,MAE)分别为370.676 MW、99.07%和246.89 MW,与单一模型和混合模型相比,实现了评价指标的明显减少。 |
中文关键词:负荷预测 主成分分析 CEEMDAN 样本熵 门控循环单元 极致梯度提升模型 |
|
Short-Term Power Load Forecasting Based on Modal Decomposition and GRU-XGBoost |
|
|
Abstract:The accurate short-term power load forecasting can effectively improve power system operations. In order to solve the problem of strong volatility and randomness of power load data affected by various factors, a load prediction method based on modal decomposition and mixed model is proposed in this paper. Firstly, the principal component analysis(PCA) method is used to process the eigenvector of the load, the redundant information is removed, and then the historical load is decomposed into simplified sub-sequences by using the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). Secondly, the sample entropy(SE) is introduced to calculate the entropy value of the sub-sequence, and the random, detailed, low-frequency and trend components of similar sub-sequences are reconstructed to obtain the random, detailed, low-frequency and trend components, and the gated recurrent unit (GRU) of different structures is selected to predict the different component types, and then the extreme gradient boosting model (XGBoost) is used to fit the residuals of each component, and the predicted value of each recombination sequence is the sum of the GRU prediction value and the XBGoost fitting value, and the final predicted value of each sequence is obtained. The results show that the root mean square error(RMSE), mean absolute percentage error(MAPE) and mean absolute error(MAE) of the proposed model are 370.676 MW, 99.07% and 246.89 MW respectively, which is significantly reduced compared with the single model and the hybrid model. |
keywords:load forecasting principal component analysis CEEMDAN sample entropy gate control loop unit extreme gradient enhancement model |
查看全文 查看/发表评论 下载PDF阅读器 |