DC Microgrid Instability Detection With Unsupervised Fault Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional stability detection methods for DC microgrids are inefficient and invasive, struggling with real-time detection and manual labeling dependencies, especially in complex multi-node systems, and lack accuracy in fault module positioning.
Innovation Solution
An unsupervised integrated method using a twinborn network framework for instability detection and fault module positioning, which includes data enhancement, feature extraction, and label mapping, enabling label-free training and precise fault identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If impedance-based methods are used to detect system stability, then the system stability analysis becomes more convenient and efficient, but disturbance signals are introduced that may generate interference to the normal operation of the system
Solution Approach 1:
The patent replaces the traditional impedance-based measurement approach (which requires physical disturbance signals) with a data-driven machine learning model that processes electrical data to predict instability. This substitution eliminates the need for actual disturbance signals while maintaining the ability to detect system stability issues.
Solution Approach 2:
The patent creates a virtual model (random forest classifier) that copies the behavior patterns of the DC microgrid system from historical data. This virtual model can predict instability without requiring actual disturbance signals to be injected into the physical system, thus avoiding interference with normal operation.
2Reliability
If conventional impedance-based methods are applied to complicated multi-node DC microgrids, then detection coverage is improved, but calculation and analysis of impedance characteristics becomes very complicated and real-time detection is difficult to achieve
Solution Approach 1:
The patent segments the complex multi-node DC microgrid system into individual converter modules, each with its own electrical data collection and analysis. The random forest model processes data from multiple segments independently and integrates results to determine overall system stability, simplifying the calculation complexity while maintaining comprehensive detection coverage.
Solution Approach 2:
The patent replaces complex impedance calculation and analysis with a machine learning-based prediction model. The random forest classifier directly processes electrical data (voltage, current, power) to predict instability, eliminating the need for complex impedance characteristic analysis while achieving real-time detection capability.
3Measurement precision
If conventional stability detection methods are used, then detection capability is achieved, but dependency on manually labeling sample tags increases which raises costs
Solution Approach 1:
The patent implements an unsupervised learning approach where the system automatically identifies instability patterns and labels data points without human intervention. The random forest model learns from unlabeled electrical data and autonomously determines which samples represent unstable conditions, eliminating the need for manual labeling while maintaining detection accuracy.
Solution Approach 2:
The patent replaces the manual labeling process with an automated machine learning system. The unsupervised random forest model automatically processes electrical data, identifies instability patterns, and creates labels without human involvement, significantly reducing labeling costs while maintaining or improving detection precision.
4Reliability
If an individual converter module breaks down in a multi-node DC microgrid, then system complexity increases for detection purposes, but fault module positioning capability is needed to maintain system reliability
Solution Approach 1:
The patent segments the DC microgrid into individual converter modules, each monitored independently for electrical parameters. When a fault occurs, the system analyzes electrical data from each segment to identify which specific module exhibits abnormal patterns, enabling precise fault positioning without requiring a complex centralized analysis system.
Solution Approach 2:
The patent implements a feedback mechanism where electrical data from each converter module continuously feeds into the random forest model. The model provides real-time feedback about the stability status of each module, enabling rapid identification of faulty modules and facilitating quick response to maintain system reliability.
Data Source
AI summary
The present invention is an unsupervised integrated method of instability detection and fault module positioning, including: acquiring a topological structure of a direct current (DC) microgrid and collecting electrical data of each node in the topological structure to construct a corresponding first enhancement dataset and a corresponding second enhancement dataset; constructing a fault type pool based on the topological structure; constructing a corresponding classification network based on a twinborn network framework; training the classification network using the prepared datasets to obtain a detection model; and inputting the electrical data of the DC microgrid to be detected to a detection model, to output whether the DC microgrid to be detected has system stability and a corresponding fault type. Further provided in the present invention is an unsupervised integrated device of instability detection and fault module positioning.