Power Grid Fault Level Mapping Using Correlated Measurements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for monitoring electric power grids lack effective means to estimate electric parameters, particularly fault levels, across different locations, especially where measurement data is unavailable or outdated, leading to uncertainties and lack of visibility in grid performance.
Innovation Solution
A method and system utilizing machine learning to correlate measurement data from one location to another, incorporating physical stimuli and electrical characteristics, allowing for the estimation of fault levels and other electric parameters across the grid, even where direct measurement data is not available, through intentional and endogenous stimuli, and forming a correlation model for real-time monitoring and prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If measurement data is collected at multiple locations across the power grid, then the coverage and visibility of grid performance improves, but the cost and complexity of the monitoring system increases
Solution Approach 1:
The patent creates a virtual copy of the measurement data by training a machine learning model (correlation model) on measurement data from one location and using it to estimate parameters at other locations. This virtual copying approach eliminates the need for physical measurement devices at every location, reducing system complexity while maintaining comprehensive grid visibility.
Solution Approach 2:
The patent introduces a machine learning correlation model as an intermediary between measurement data from one location and the estimated parameters at other locations. This intermediary processes and transforms the data, enabling indirect measurement across the grid without requiring direct physical presence at all measurement points.
2Measurement precision
If measurement devices are deployed at all grid locations, then the accuracy of local parameter measurement improves, but the cost and complexity of the system increases
Solution Approach 1:
The patent uses machine learning to create accurate virtual copies of measurement data at locations where physical devices are not present. The correlation model learns the relationships between measurements at different locations and reproduces accurate estimates of fault levels and other parameters, achieving measurement precision without deploying physical devices everywhere.
Solution Approach 2:
The patent segments the monitoring function into two parts: physical measurement at selected locations and virtual estimation at other locations through machine learning. This segmentation allows the system to achieve comprehensive monitoring coverage while using fewer physical devices, reducing complexity while maintaining accuracy through the combination of actual measurements and AI-based estimations.
3Quantity of substance
If physical stimuli are intentionally generated to obtain measurement data, then the availability of training data for correlation modeling improves, but the disturbance to the power grid increases
Solution Approach 1:
The patent employs periodic or intermittent physical stimuli rather than continuous disturbances. By applying stimuli at specific intervals and using the machine learning model to interpolate between stimulus events, the system accumulates sufficient training data while minimizing the overall disturbance to the power grid operation.
Solution Approach 2:
The patent uses partial physical stimuli that are sufficient to generate measurable responses for training the correlation model, but not excessive enough to cause significant grid disturbance. The machine learning model is trained on these partial stimulus responses and then used to predict parameters under normal operating conditions without requiring continuous strong stimuli.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
This document discloses a solution for a method of monitoring an electric power grid. According to an aspect, a method comprises: detecting one or more physical stimuli in the electric power grid; obtaining, while the one or more physical stimuli is effective, a first set of measurement data associated with a first location of the electric power grid; computing operational information such as a fault level of the first location of the electric power grid on the basis of the first set of measurement data; mapping the operational information to corresponding operational information of a second location of the electric power grid on the basis of the first set of measurement data and correlation between electrical characteristics of the first location and the second location.