PMU bad data detection method based on spectral clustering

Abstract

The invention discloses a PMU bad data detection method based on spectral clustering. The method comprises the steps: building a decision tree model based on the slope features of four-point data, and recognizing event data, normal data and bad data through the decision tree model; for the identified data set A containing normal data and bad data, a 3 sigma criterion is used for preliminary screening, the data set is divided into an A1 part, an A2 part and an A3 part, the A1 part and the A2 part are the normal data and the bad data respectively, and the A3 part contains the normal data and the bad data; and constructing a weight distance matrix between the data by using a spectral clustering method so as to detect bad data in the A3 part. According to the method, bad data with a small deviation value can be accurately detected through the weight between the data.

Description

[0001] This application claims the priority of the patent application 202011576078.X filed on 2020-12-28. technical field [0002] The invention relates to the technical field of power systems, in particular to a method for detecting PMU bad data based on spectral clustering. Background technique [0003] PMU can provide real-time phasor data for various applications in power systems, such as decision-making control, oscillation detection, and state estimation. Data quality issues of varying degrees. Accurate detection of PMU bad data is crucial to improving data quality and ensuring safe and stable operation of power systems. At present, the commonly used methods for detecting PMU bad data are state estimation-based, Kalman filter-based and data-driven methods. Among these methods, data-driven methods have attracted extensive attention because they do not require prior knowledge of system topology and line parameters. [0004] Existing data-driven methods are based on alg...

AI Technical Summary

Problems solved by technology

[0004] Existing data-driven methods are based on algorithms such as low-rank, principal component analysis, and spatio-temporal similarity. However, they all require the measurement information of multiple PMUs. For some areas, only a small number of PMUs are installed, and it is difficult to obtain The measurement information of multiple PMUs is not applicable; and the methods of using the measurement of a single PMU to detect bad data are based on ensemble learning and density clustering. However, when bad data appears during the event These methods may not be applicable when

Images