Harmonic anomaly identification method based on variable point segmentation and sequence clustering

Abstract

The invention relates to a harmonic anomaly identification method based on variable point segmentation and sequence clustering. A change point of harmonic monitoring statistical data in a long-term operation process is detected; the monitoring data sequence is segmented through each local variable point of the data to reflect harmonic state change points under an actual system, clustering based ona distance function for measuring form similarity is carried out on each section of sequence to obtain a form type of each section of sequence, so that identification of abnormal types is realized, and subsequent analysis of causes and governance of abnormal mechanisms are facilitated.

Description

technical field [0001] The invention relates to a harmonic abnormality identification method based on change point segmentation and sequence clustering. Background technique [0002] With the advancement of grid power electronics, a large number of non-linear loads are connected to the grid, and applications such as distributed power grid connection and energy storage are increasing, which pose challenges to high-quality power quality. Among them, the harmonic problem is a kind of serious steady-state power quality problem. Monitoring and evaluating the harmonics of the power grid and discovering abnormal harmonics in time is an important basis for controlling and improving harmonic problems and providing high-quality power supply. [0003] Due to the continuous advancement of the construction of the grid power quality monitoring system, the arrangement of monitoring terminals has increased, and a large amount of harmonic monitoring statistics have been stored, providing a ...

AI Technical Summary

Problems solved by technology

Among them, the harmonic problem is a kind of serious steady-state power quality problem.

The main problem of these two types of methods is that the statistical features selected by the statistical model or the clustering process do not include the trend change characteristics of the data. In addition, the fixed statistical model is difficult to apply to the data of all monitoring points.

On the other hand, some studies take into account the morphological similarity between data, and use the degree of similarity with normal conditions for abnormal judgment and early warning. However, the problem with this type of method is only for periodic harmonics that change regularly. An abnormality identification method using a fixed analysis time window

In the case where the common connection point is affected by the interaction of different harmonic interference sources, and the periodic regularity of the harmonic change characteristics over a long period of time is not obvious, the analysis method with a fixed time window has limitations

Images