基于多阶段数据递推分析的用户用电行为特性挖掘方法  

User behavior characteristic mining method based on multi-stagedata recurrence analysis

在线阅读下载全文

作  者:李延珍 王海鑫[1] 杨子豪 马一鸣 杨俊友[1] 陈哲[2] LI Yanzhen;WANG Haixin;YANG Zihao;MA Yiming;YANG Junyou;CHEN Zhe(School of Electrical Engineering,Shenyang University of Technology,Shenyang 110870,China;Department of Energy Technology,Aalborg University,Aalborg DK-9220,Denmark)

机构地区:[1]沈阳工业大学电气工程学院,辽宁沈阳110780 [2]丹麦奥尔堡大学能源技术系,奥尔堡DK-9220

出  处:《电机与控制学报》2025年第2期35-46,共12页Electric Machines and Control

基  金:高等学校学科创新引智计划资助项目(D23005);辽宁省教育厅基本科研项目(JYTQN2023288);辽宁省科技计划联合计划项目(2023JH2/101700275)。

摘  要:用户用电行为特性分析为构建居民用户的家庭智能用电策略提供了可靠的理论基础和数据支撑。为提取有效直观负荷用电行为特征,提出一种基于多阶段数据递推分析的用户用电行为特性挖掘方法。该方法第一阶段构建了基于混合深度学习的非侵入式负荷分解模型,将用户集中数据分解为电器设备用电数据分量集合;第二阶段提出了基于卡尔曼滤波与广义似然比检验的事件检测方法,对电器设备的启停状态进行了判定;第三阶段量化用户用电行为特性,并提出了基于核密度估计的电器设备差异化时域概率模型。以公开数据UK-DALE为对象展开仿真验证,实验结果表明,该方法能有效捕捉用户细粒度能耗数据,构建智能电表与用户用电特性之间的桥梁,为优化管理及集群调控用户负荷提供有效手段。The analysis of electricity consumption characteristics provides a reliable theoretical basis and data support for constructing a new demand response strategy for residential users.This paper proposes a new method for extracting the effective and intuitive user behavior characteristics based on multi-stage data recursive analysis.In the first stage,a hybrid deep learning disaggregation model based on non-intrusive load monitoring is constructed,and the centralized data of users is decomposed into the data components of electrical appliances.In the second stage,an event detection method based on Kalman filtering and a generalized likelihood ratio test is proposed to determine the start-stop state of electrical appliances.In the third stage,the load characteristics are quantified,and the time-domain probability model of differentiation of electrical appliances based on kernel density estimation is proposed.The open data set UK-DALE is utilized for simulation verification.The experimental results show that the proposed method can effectively capture the fine-grained energy consumption data of users,build a bridge between smart meters and users'power consumption characteristics,and provide an effective means for load optimization management and cluster regulation.

关 键 词:非侵入式负荷监测 深度学习 用电特性分析 事件检测 多阶段递推 核密度估计 

分 类 号:TM714[电气工程—电力系统及自动化]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象