从用户侧电流数据中发现用电事件,对挖掘用户用电行为模式,提高用户侧用电管理水平具有重要意义.为及时有效地检测出单个电器下电流数据中蕴含的用户用电事件,设计基于聚类用户用电事件辨识模型.该模型在用户用电电流数据高频在线监测基础上,构建固定宽度电流序列片段,将电流序列中用电事件辨识问题视为电流序列片段集的聚类划分问题,并使用轮廓系数和精度2个指标进行性能评估.实验结果表明,相较基于k均值聚类、层次式聚类以及SOM(Self-Organizing Map)聚类等实现的用户用电事件辨识模型,基于DBSCAN(Density-Based Spatial Clustering of Applications with Noise)算法用户用电事件辨识模型可以高效辨识出高频电流序列中的用户用电事件.
An Approach to Model for the Identification of Electrical Event Based on Clustering Analysis
The identification of electric events from user-side electric current data is of great importance in exploring patterns of user electric consumption behavior and enhancing user-side electric management capabilities.To timely and effectively detect user electric events contained in the electric data under a single electrical appliance,a model for electric event identification based on clustering analysis is proposed.Building upon frequent online monitoring of user electric current data,this model constructs fixed-width segments of the current sequence,treats the identification problem of electric events in the current sequence as a clustering partitioning problem of the segment set,and evaluates the performance using two metrics,i.e,silhouette coefficient and precision.The experiments demonstrate that,compared to the identification models of electric events implemented based on k-means clustering,hierarchical clustering,and SOM(Self-Organizing Map)clustering,the identification model of electric events based on DBSCAN(Density-Based Spatial Clustering of Applications with Noise)can be more effective for the identification of user electrical events.