Power Line Disturbance Classification Using RMS Difference Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for detecting and classifying disturbance conditions in power transmission lines, such as electrical faults and power swings, are ineffective near swing center points or in three-phase transmission lines, requiring complex settings and high-quality phasor estimates.
Innovation Solution
A method utilizing an Intelligent Electronic Device (IED) that employs a machine learning technique, specifically an extreme gradient boost classification model, to process difference values of electrical parameter magnitudes (RMS values) measured in power transmission lines, enabling the detection and classification of load changes, power swings, and electrical faults.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional techniques (blinders, SCV technique, resistance technique) are used to detect and differentiate fault disturbances, then normal faults can be differentiated from power swings, but these techniques are not effective for disturbances near swing center points or in three phase transmission lines and require complex system specific settings
Solution Approach 1:
The patent replaces conventional mechanical/mathematical techniques (blinders, SCV technique, resistance technique) with a machine learning-based classification system. The IED uses trained classification models to automatically distinguish between fault conditions and power swing conditions based on electrical parameter patterns, eliminating the need for complex system-specific settings and manual configuration while maintaining high detection accuracy even near swing center points.
Solution Approach 2:
The patent changes the approach from using fixed threshold-based parameters to using adaptive machine learning models that learn optimal parameter relationships from training data. The system measures electrical parameters (voltage, current, power) and uses classification algorithms to detect disturbance types, allowing the system to adapt to different grid conditions without requiring manual reconfiguration of complex settings.
2Reliability
If elaborate procedures requiring large computations are used to detect disturbances in low inertia power systems with renewable energy sources, then disturbance detection can be achieved, but the computational burden increases significantly
Solution Approach 1:
The patent applies partial action by implementing a two-stage detection approach: first using simplified real-time measurements and quick classification for immediate disturbance detection, and only invoking more complex analysis when initially needed. The system uses lightweight classification models that provide sufficient accuracy for real-time operation without requiring exhaustive computational procedures, thus reducing energy consumption while maintaining reliable disturbance detection capability.
3Reliability
If protection relay is operated during fault condition to disconnect part of the grid, then equipment is protected, but power swing conditions can also operate the protection relay causing blackout
Solution Approach 1:
The patent implements feedback-based classification where the machine learning model continuously analyzes electrical parameter patterns and provides feedback to the protection system. The classification model learns from training data the distinctive patterns of fault conditions versus power swing conditions, enabling it to provide accurate classification feedback that prevents false operation of protection relays during power swings while maintaining proper operation during actual faults, thus eliminating false alarms that cause blackouts.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present specification provides a method and device for determining a disturbance condition in a power transmission line. The method includes obtaining (302) a plurality of sample values corresponding to an electrical parameter measured in each phase. The method further includes determining (304) a plurality of magnitudes of the electrical parameter corresponding to each phase based on the corresponding plurality of sample values and determining (306) a plurality of difference values for each phase based on the corresponding plurality of magnitudes. The method includes processing (308) the plurality of difference values using a machine learning technique to determine the disturbance condition. The disturbance condition is one of a load change condition, a power swing condition and an electrical fault condition. The method also includes performing (310) at least one of a protection function and a control function based on the disturbance condition.