摘要
配电网作为电力系统的关键环节,有必要识别配电网潜在危害,避免失稳。为了解决数据中噪声干扰的问题并提高态势预测准确性,提出了一种基于深度学习的配电网安全态势感知方法。首先,采集配电网运行量,利用奇异值分解(singular value decomposition,SVD)对运行量进行降噪;其次,分析运行量与安全态势的关系,采用评估值指标评估配电网态势;最后,利用注意力时域卷积网络(temporal convolution network-attention mechanism,TCNAM)对降噪后的输入数据预测得出态势评估值,预测配电网潜在危害,若失稳,则发出预警信号。通过对IEEE 33节点系统和实际配电网系统仿真可知,TCN-AM预测效果好,且进行降噪处理后预测准确性有所提高,能够在满足预警条件后,发出相应的预警信号。所提方法在降噪处理后能够更准确地实现配电网的安全态势感知。
As the key link of the power system,it is necessary to identify the potential hazards of the distribution network and avoid the instability.To solve the problem of noise interference and improve the accuracy of situa⁃tion prediction,a security situation awareness method of the distribution network based on deep learning is stud⁃ied.Firstly,The operating quantity of the distribution network is collected and denoised by singular value de⁃composition(SVD).Secondly,the relationship between operation volume and security situation is analyzed,and the situation of the distribution network is evaluated by evaluating the value index.Finally,the temporalconvolution network-attention mechanism(TCN-AM)is used to predict the input data after noise reduction to obtain the situation assessment value,predict the potential hazards of the distribution Network,and send an ear⁃ly warning signal if it is unstable.Through the simulation of the IEEE 33-bus system and actual distribution net⁃work system,TCN-AM has a good prediction effect,and the prediction accuracy is improved after noise reduc⁃tion processing,and the corresponding early warning signal can be issued after meeting the early warning condi⁃tions.The proposed method can realize the security situation awareness of the distribution network more accu⁃rately after noise reduction.
作者
管委玲
何金波
陈潞
倪烨
杨军
裘炜望
李赟
GUAN Weiling;HE Jinbo;CHEN Lu;NI Ye;YANG Jun;QIU Weiwang;LI Yun(Taizhou Luqiao District Power Supply Company,State Grid Zhejiang Electric Power Company,Taizhou 318050,China;Luqiao Branch,Zhejiang Taizhou Hongtai Power Supply Service Company,Taizhou 318050,China)
出处
《电力学报》
2024年第1期11-20,共10页
Journal of Electric Power
基金
国家自然科学基金(52007103)
国网浙江省电力有限公司台州市路桥区供电公司基于深度学习的配电网安全态势感知研究(5211T223000M)。
