基于数据驱动的电力系统异常事件检测研究

Research on Abnormal Event Detection of Power System Based on Data Driven

为提高电力系统异常事件检测的有效性,提出了一种基于数据驱动的电力异常检测模型。为减轻通信系统的数据传输负担,设计一种不等间隔压缩与重构方法。为测量任意母线之间动态特性的综合相似性,利用基于主成分分析的相似性搜索方法来处理PMU数据之间的相关性。同时,基于可达距离对某一区域电力系统中的异常事件进行有效检测。以某省电力系统PMU记录的数据为例,对所提模型进行验证。结果表明,所提模型能够有效对数据进行压缩,压缩后的PMU数据与原始PMU数据非常吻合,且数据大约压缩到原始规模的10%,可大幅度减轻通信系统的负担。与等间隔压缩法相比,所提不等间隔压缩方法重构误差明显较低,鲁棒性更优。

In order to improve the effectiveness of power system anomaly detection,a data driven power anomaly detection model is proposed.In order to reduce the data transmission burden of communication system,a method of unequal interval compression and reconstruction is designed.In order to measure the comprehensive similarity of dynamic characteristics between any buses,the similarity search method based on principal component analysis is used to deal with the correlation between PMU data.At the same time,based on the reachable distance,the abnormal events in the power system of a certain region can be effectively detected.Taking the data recorded by a provincial power system PMU as an example,the proposed model is verified.The results show that the proposed model can effectively compress the data,and the compressed PMU data is very consistent with the original PMU data,and the data is compressed to about 10%of the original size,which can greatly reduce the burden of the communication system.In addition,compared with the equal interval compression method,the proposed unequal interval compression method has lower reconstruction error and better robustness.