Power Network State Estimation Error Localization by Recursive Splitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing state estimation methods in power systems are prone to errors in static and dynamic data, leading to biased solutions or non-convergence, which can be exacerbated by incorrect assumptions in network topology and measurement data.
Innovation Solution
A recursive network splitting method is employed to identify non-convergent portions of the network model, followed by detailed inspection to pinpoint plausible data errors, using spectral factorization and modified state estimation algorithms to isolate and correct errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If the network model is split into multiple portions, then errors can be localized and identified, but the complexity of the detection process increases
Solution Approach 1:
The network model is divided into multiple portions, enabling errors to be localized to specific segments rather than the entire network. This segmentation approach systematically reduces the search space for errors while maintaining a manageable detection process through structured execution of the state estimation algorithm on each portion.
Solution Approach 2:
The state estimation algorithm execution provides feedback on convergence status and solution quality for each network portion. This feedback mechanism identifies which portions contain errors, guiding the localization process without requiring complex external analysis tools or procedures.
2Measurement precision
If detailed inspection is performed on non-convergent portions, then plausible data errors can be identified, but the time required for error detection increases
Solution Approach 1:
By dividing the network into portions and executing state estimation on each segment, the inspection process focuses only on non-convergent portions rather than the entire network. This targeted approach maintains high error identification accuracy while reducing overall detection time by excluding already-validated network segments.
Solution Approach 2:
The state estimation algorithm is executed on network portions before detailed error inspection, preliminarily identifying which portions are non-convergent and likely contain errors. This preliminary action filters out error-free portions, so detailed inspection is only performed on problematic segments, reducing total detection time.
Data Source
Figure 1~2
Figure 3~4
Figure 5~8
AI summary
Methods for detecting network model data errors are disclosed. In some examples, methods for detecting network model data errors may include splitting a network model into a first plurality of portions, executing an algorithm on each of the portions, identifying a portion for which the algorithm is determined to be non- converged, splitting the identified portion into a second plurality of portions, repeating the executing, identifying and splitting the identified portion until a resulting identified portion is smaller than a predefined threshold, and examining the resulting identified portion to identify plausible data errors therein. In some examples, examining the resulting identified portion to identify plausible data errors therein may include executing a modified algorithm, which may include an augmented measurement set, on the identified portion.