摘要
传统配电变压器重过载预警方法主要基于确定性的预测,容易忽视由于负荷不确定性带来的风险影响。针对该问题,提出计及负荷不确定性的配电变压器重过载风险预警方法。首先,采用门控循环单元分位数回归算法预测配电变压器在不同分位点上的负荷水平,并结合核密度估计方法进一步得到未来负荷的概率密度函数,以提供反映负荷不确定性的预测信息;其次,通过效用函数描述配电变压器承受重过载事故的严重程度,结合电力系统风险理论,评估配电变压器可能面临的重过载风险水平;最后,根据预先定义好的标准划分重过载风险等级,进而生成涵盖风险详情的告警信息。结合广东省某地区配电网展开具体的算例分析,验证了所提方法的可行性和有效性。
The traditional heavy overload early warning method for distribution transformers is mainly based on deterministic predictions,and it is easy to ignore the risk impact caused by load uncertainty.To solve this problem,an early warning method of heavy overload risk for distribution transformers,taking into account the load uncertainty,is proposed in this paper.First,the quantile regression gated recurrent unit is used to predict the load level of distribution transformers at different quantile points,and combined with the kernel density estimation method to further obtain the probability density function of the future load to provide the forecast information that reflects the uncertainty of the load.Second,the utility function is used to describe the severity of the heavy overload accident of distribution transformers,and combined with the power system risk theory to evaluate the possible heavy overload risk level of distribution transformers.Finally,the heavy overload risk level is divided according to the predefined standard,and then the warning information covering the details of the risk is generated.Based on a specific example analysis of the distribution network in a certain area of Guangdong Province,the feasibility and effectiveness of the proposed method are verified.
作者
黄园芳
刘云凯
郑世明
李旺军
彭显刚
林泽鑫
HUANG Yuanfang;LIU Yunkai;ZHENG Shiming;LI Wangjun;PENG Xiangang;LIN Zexin(Zhanjiang Power Supply Bureau of Guangdong Power Grid Corporation,Zhanjiang 524005,Guangdong,China;School of Automation,Guangdong University of Technology,Guangzhou 510006,Guangdong,China)
出处
《电网与清洁能源》
北大核心
2021年第10期17-24,共8页
Power System and Clean Energy
基金
广东电网有限责任公司科技项目(030800KK52180032)。
关键词
配电变压器
重过载风险预警
负荷概率预测
不确定性分析
深度学习
distribution transformer
heavy overload risk early warning
load probability prediction
uncertainty analysis
deep learning
