摘要
针对配电台区负荷预测精度低的问题,提出一种结合变分模态分解(VMD)、迭代自组织数据分析算法(ISODATA)与深度信念网络(DBN)的短期负荷预测方法。首先,利用VMD将原始负荷序列分解成多个模态分量。然后,采用ISODATA对特征相似的各分量进行聚类。最后,对每类分量分别建立DBN预测模型,并叠加相应预测结果实现负荷预测。算例分析结果表明,相较于其他方法,所提方法有效提高了电力负荷预测的效率和精度,具有很大的应用潜力。
Targeting the problem of low accuracy of load prediction in distribution station area,the paper proposes a short-term combined load forecasting method based on variational mode decomposition(VMD),iterative self-organizing data analysis techniques algorithm(ISODATA)and deep belief network(DBN).Firstly,VMD is used to decompose the original load series into several modal components.Then ISODATA is applied to cluster the above components,and the components with similar load characteristics are grouped into one group.Finally,a DBN model is used to predict every group of the components separately,and the prediction results are overlapped to achieve the load prediction.The results show that the proposed method has higher prediction accuracy and shows a potential in real application compared to other methods.
作者
寿绍安
罗海荣
王晓康
张洁
虎俊
周剑桥
SHOU Shao’an;LUO Hairong;WANG Xiaokang;ZHANG Jie;HU Jun;ZHOU Jianqiao(State Grid Wuzhong Power Supply Company,Wuzhong 751100,China;State Grid Ningxia Electric Power Research Institute,Yinchuan 750000,China;Shanghai Jiaotong University,Shanghai 201100,China)
出处
《智慧电力》
北大核心
2023年第11期53-60,共8页
Smart Power
基金
国家自然科学基金资助项目(52107200)
国网宁夏电力有限公司科技项目(5229W220007)。
