PMU Bad Data Detection via Spectral Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Phasor measurement units (PMUs) face data quality issues due to interference and synchronization jitter, leading to compromised data that can affect the safety and stability of power systems, with existing methods requiring system topology and parameters, which can result in misjudgment and inefficiency in bad data detection.
Innovation Solution
A data-driven PMU bad data detection algorithm based on spectral clustering and decision trees, which identifies event data and bad data without relying on system topology or parameters, using the slope feature of data and spectral clustering to detect bad data with small deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing bad data detection methods are used, then detection capability is provided, but system topology and parameters are required, leading to misjudgment and inefficiency
Solution Approach 1:
The patent extracts and removes the dependency on system topology and parameters from the bad data detection process. By using a data-driven approach with spectral clustering, the method isolates the detection capability from complex system information requirements, achieving accurate detection without needing topology data or system parameters.
Solution Approach 2:
The detection algorithm performs self-service by automatically identifying bad data through spectral clustering of the measurement data itself. The system uses the inherent characteristics of the data (eigenvectors and eigenvalues) to detect anomalies without external system information, making the detection process autonomous and efficient.
2Measurement precision
If spectral clustering is used for bad data detection, then detection of small deviation bad data is achieved, but computational complexity increases
Solution Approach 1:
The patent segments the detection process into distinct stages: data collection, spectral decomposition to obtain eigenvectors and eigenvalues, clustering analysis, and bad data identification. This segmentation allows the complex spectral clustering computation to be organized efficiently, managing computational load while maintaining high precision in detecting small deviation bad data.
3Ease of operation
If data-driven method is used, then system topology and parameters are not required, but detection of event data and bad data differentiation is needed
Solution Approach 1:
The patent implements feedback mechanisms where the spectral clustering results are analyzed to distinguish between event data and bad data. The algorithm uses the clustering outcomes and deviation patterns as feedback to refine detection decisions, enabling differentiation without requiring system topology or parameters, thus maintaining ease of operation while resolving the detection difficulty.
Data Source
AI summary
A data-driven PMU bad data detection algorithm based on spectral clustering using single PMU data is described. The described algorithm does not require the system topology and parameters. First, a data identification method based on a decision tree is described to distinguish event data and bad data by using the slope feature of each set of data. Then, a bad data detection method based on spectral clustering is described. By analyzing the weighted relationships among all the data, this method can detect the bad data that has a small deviation.