基于停电事件的户变异常研判方法研究

Research on the Method of Household Abnormity Analysis Based on Power Outage Event

当前低压配电网数据采集不全面,配电台区拓扑关系存在不确定性,为了解决这个问题,研究设计配电台区的户变异常识别系统。系统的户变关系异常识别模型中由BP神经网络和SOM神经网络构成,完成数据的聚类分析和网络映射,基于提取出的特征信息完成异常识别任务。系统的拓扑识别模块中加入了超级电容组,在失电状态下仍能够完成用户侧信息的采集和停电事故的上报。实验结果显示,该研究系统中的识别效率最高,识别时间最低为2724 ms,异常识别最高为98.7%。

In view of the incomplete data collection of the current low-voltage distribution network and the uncertainty of the topological relationship of the distribution platform,this study designs a household variation anomaly recognition system for the distribution platform.The anomaly recognition model of household variation relationship in the system is composed of BP neural network and SOM neural network,which completes the clustering analysis and network mapping of data,and completes the anomaly recognition task based on the extracted feature information.A supercapacitor group is added to the topology identification module of the system,which can collect user information and report power outage accidents even when the system is powered off.Experimental results show that the system has the highest recognition efficiency,the lowest recognition time is 2724 ms,and the highest anomaly recognition is 98.7%.