电力现货市场中用户电量异常数据辨识方法研究
Research on Identification Methods for Abnormal User Electricity Data in the Electricity Spot Market
何妍妍 1赵志扬 1程叙鹏 1陈奕汝 1吴秀英1
作者信息
- 1. 国网浙江省电力有限公司营销服务中心,浙江杭州 310000
- 折叠
摘要
针对现有辨识方法对用户用电量异常数据映射越限不明显及辨识率较低的问题,提出电力现货市场中用户电量异常数据辨识方法.通过采集用户用电异常数据,对缺失值进行补充和归一化处理,计算当前用户用电量数据和历史用电量间的差异,再建立模型观测用电量与系统状态之间的关系,获取不同时间段内的变化规律.运用非参数密度法统计电量数据,提取日电量特征及加权后的用户特征曲线和用户电量数据的可行矩阵,比较待辨识数据曲线的上限、下限是否在可行范围内,从而完成异常数据的辨识.结果表明,电量异常数据映射至可行域中存在明显超越可行域上限、下限的情况,实验组的异常数据辨识准确率为100%,实现了电力现货市场中对用户电量异常数据的精准辨识.
Abstract
A method for identifying abnormal user electricity consumption data in the electricity spot market is proposed to address the issues of unclear mapping of abnormal user electricity consumption data and low identification rate using existing identification methods.By collecting abnormal user electricity consumption data,supplementing and normalizing missing values,calculating the difference between current user electricity consumption data and historical electricity consumption,and then establishing a model to observe the relationship between electricity consumption and system status,obtaining changes in different time periods.Using the non parametric density method to statistically analyze electricity consumption data,extracting daily electricity consumption characteristics,weighted user characteristic curves,and feasible matrices of user electricity consumption data,comparing whether the upper and lower limits of the data curve to be identified are within the feasible range,in order to complete the identification of abnormal data.The results showed that there were significant cases of exceeding the upper and lower limits of the feasible domain when mapping abnormal electricity data to the feasible domain.The experimental group's accuracy in identifying abnormal data was 100%,achieving accurate identification of user abnormal electricity data in the electricity spot market.
关键词
电力现货市场/用户电量/异常数据/信息采集/辨识模型Key words
electricity spot market/user electricity consumption/abnormal data/information collection/identification model引用本文复制引用
出版年
2024
NETL
NSTL
万方数据