基于多源异构数据关联规则分析的需求响应潜力居民用户辨识方法

Recognition of Demand Response Potential Resident Users Based on Multi-source Heterogeneous Data Association Rule Analysis

精准辨识需求响应潜力居民用户群体,形成一定规模的聚合资源参与电网互动可以进一步发挥需求响应调节能力。针对当前辨识研究仅采用负荷数据挖掘,难以精准识别需求响应潜力居民用户的不足,提出了一种基于多源异构数据关联规则分析的潜力用户辨识方法。首先,针对负荷形态辨识中负荷类别划分精细化程度较低的问题,采用双向门控循环单元网络针对用户典型日负荷曲线进行分类,并依据各类用电负荷峰谷幅值及时段分布特征,实现需求响应潜力用户的初步选择。其次,针对用户潜力评估指标维度单一、指标权重同质化的不足,融合多源异构数据进行关联规则分析,构建多维需求响应潜力评估指标体系,并基于相关性与冗余性分析赋予指标客观权重,利用逼近理想解排序算法求解用户潜力值排序。最后,通过算例验证了所提方法的有效性。

Accurately recognizing potential resident user groups for demand response to form a certain scale of aggregated resources to participate in the power grid interaction may further develop the adjustment ability of demand response.In view of the deficiency that the current recognition researches only use load data for mining,which is difficult to accurately recognize the potential resident user groups for demand response,a recognition method based on the multi-source heterogeneous data association rule analysis is proposed.Firstly,aiming at the problem of low refinement of the load classification in the load pattern identification,a classification method based on Bi-GRU network is proposed to realize the preliminary selection of the demand response potential users according to the peak and valley amplitudes and the time distribution characteristics of various kinds of loads;Secondly,in view of the shortcomings of one-sided dimension of the user potential evaluation indicators and the homogenization of indicator weights,the association rules are analyzed by integrating the multi-source heterogeneous data.The multi-dimensional indicator system for the demand response potential evaluation is constructed.The weights of the indicators are extracted based on the correlation and redundancy analysis.The potential values of the users are sorted by using the TOPSIS.Finally,an example is given to verify the effectiveness of the proposed method.