Automatic Power Topology Discovery Using Bayesian Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial environments often lack accurate electrical network topologies, making it difficult to locate power disturbance events, infer causes, and maintain systems due to the complexity and time-consuming nature of manually updating power topologies, especially in legacy systems with many changes.
Innovation Solution
A method for automatic electrical network topology discovery using power data from devices within a network, detecting power change events, and updating probabilities of connections between devices based on Bayesian methods to generate accurate topology diagrams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual methods are used to maintain power topology diagrams, then accuracy can be maintained through human review, but the process becomes extremely time-consuming and difficult to keep updated
Solution Approach 1:
The system automatically discovers and updates power topology by monitoring power data from devices themselves, eliminating the need for manual diagram maintenance. Devices self-report their power consumption changes, and the system autonomously reconstructs topology based on this data, making the system self-updating without human intervention.
Solution Approach 2:
The patent replaces manual mechanical processes (humans drawing and updating diagrams) with an automated computational system that uses power data analysis and probabilistic algorithms to discover and maintain topology information automatically.
2Measurement precision
If traditional topology identification methods are used, then power distribution side topology can be identified, but lower level load branch topology cannot be detected
Solution Approach 1:
The system segments the power network into different hierarchical levels (power distribution side and load branches) and applies targeted analysis methods to each segment. By dividing the monitoring scope into manageable segments, the system can effectively identify topology at both the high-level distribution and low-level branch configurations.
Solution Approach 2:
The patent extends topology identification from the traditional single dimension (power distribution side) to multiple dimensions by incorporating load branch level analysis. This multi-dimensional approach allows simultaneous identification of both upstream distribution topology and downstream load connections.
3Measurement precision
If complete inventory of installed equipment is required, then accurate topology can be established, but the complexity and time required increases significantly
Solution Approach 1:
Instead of requiring complete physical inventories and manual documentation, the system creates a dynamic digital copy of the power topology by analyzing power data patterns. This virtual model is continuously updated based on observed power consumption changes, providing accurate topology information without the complexity of maintaining complete equipment inventories.
Solution Approach 2:
The system transforms the approach from static inventory-based topology establishment to dynamic parameter-based discovery. By monitoring changing parameters (power consumption, timing patterns) rather than relying on fixed inventory data, the system achieves accurate topology with reduced complexity.
Data Source
AI summary
Various embodiments of the present technology generally relate to power topology discovery in industrial environments. More specifically, some embodiments relate to automatic power topology discovery for factories based on device data that is already recorded for other purposes. Systems and methods described herein may be used to generate an accurate electrical network topology by collecting power data from power devices that may provide real-time or recorded measurements, detecting power change events, and matching power change signatures over power events for the devices in order to calculate the likelihoods of possible topology assumptions. Power change event data is used to recursively update topology probabilities using the Bayesian formula until a system topology can be produced with satisfactory confidence.