基于重组二次分解及LSTNet-Atten的短期负荷预测  

Short term load forecasting based on recombination quadratic decomposition and LSTNet-Atten

作　　者：刘洪伟 王磊 刘阳 张鹏超[2] 乔石 LIU Hongwei;WANG Lei;LIU Yang;ZHANG Pengchao;QIAO Shi(School of Electrical Engineering,Shanxi University of Technology,Hanzhong 723001,China;Key Laboratory of Industrial Automation,Shanxi University of Technology,Hanzhong 723001,China;Jinzhong Power Supply Company,State Grid Shanxi Electric Power Company,Jinzhong 030600,China)

机构地区：[1]陕西理工大学电气工程学院,陕西汉中723001 [2]陕西理工大学工业自动化重点实验室,陕西汉中723001 [3]国网山西省电力公司晋中供电公司,山西晋中030600

出　　处：《浙江大学学报(工学版)》2025年第5期1051-1062,共12页Journal of Zhejiang University:Engineering Science

基　　金：国家自然科学基金:资助项目(62176146);国家社会科学基金:西部项目(21XTY012);陕西理工大学研究生创新基金:资助项目(SLGYCX2405)。

摘　　要：针对电力负荷数据随机性强、波动性大,预测精度较低的问题,提出基于重组二次分解及LSTNet-Atten的短期负荷预测方法.在数据预处理阶段,采用自适应白噪声的完全集合经验模态分解对负荷序列进行初步分解,降低原始信号的随机性和波动性.根据子序列的样本熵值,将相似的子序列重组聚合.在特征工程阶段,采用变分模态分解对重组得到的复杂度较高的分量进行再次分解,通过皮尔逊、斯皮尔曼、最大信息系数方法评估输入影响因素与负荷数据之间的相关性,利用证据理论优化输入数据的特征维度.在模型构建阶段,重构LSTNet-Atten预测模型,采用卷积模块挖掘序列的局部依赖关系,通过循环和循环跳过模块提取数据的长短期特征,提高数据本身的可预测性.利用自回归模块增强神经网络对线性特征的识别能力,提高模型的预测性能.增加时间注意力赋予重要特征更多的权重,实现全局与局部联系的捕获.在瓦伦西亚区域级负荷数据集上的实验结果表明,与其他经典的深度学习模型相比,所提方法的序列预测误差最高降低了66.69%,拟合系数提高了5.04%,预测精度和鲁棒性更高.A short-term load forecasting method based on restructured second modal decomposition and LSTNet-Atten was proposed in order to address the issues of high randomness and volatility in power load data,as well as the relatively low prediction accuracy.Complete ensemble empirical mode decomposition with adaptive white noise was employed for the preliminary decomposition of the load sequence during the data preprocessing phase.Then the randomness and volatility of the original signal was reduced.Similar subsequences were reorganized and aggregated based on their sample entropy values.Variational modal decomposition was applied to further decompose the components with higher complexity obtained from the reorganization in the feature engineering phase.The correlation between input influencing factors and load data was evaluated by using Pearson,Spearman,and maximum information coefficient methods,while evidence theory was utilized to optimize the feature dimensions of the input data.The LSTNet-Atten forecasting model was reconstructed in the model construction phase.Convolutional modules were used to mine local dependencies within sequences,while recurrent and skip recurrent modules extract both long-term and short-term features from the data to enhance its predictability.The autoregressive module was used to enhance the ability of the neural network to recognize linear features and improve the predictive performance of the model.More weight was given to important features by increasing temporal attention in order to achieve the capture of global and local connections.Experimental results on a Valencia regional load dataset indicate that the proposed method reduces sequence prediction error by up to 66.69%compared with other classical deep learning models with a 5.04%increase in fitting coefficient,demonstrating higher prediction accuracy and robustness.

关 键 词：短期负荷预测 二次分解 样本熵 LSTNet 证据理论 敏感特征因子筛选 注意力机制 

分 类 号：TM715[电气工程—电力系统及自动化]