Power Supply Network State Modeling Under Metering Uncertainty
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Low-voltage power supply networks lack accurate metering infrastructure, leading to imprecise state estimation and reliance on deterministic methods that do not account for uncertainty, making it difficult to manage changes in generation and consumption scenarios.
Innovation Solution
A method using GIS data to create a branch model, combined with system data, followed by load flow simulation and Bayesian neural networks to determine a state model that incorporates probabilistic weights, allowing for real-time state estimation and control interventions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deterministic state estimation methods are used in low-voltage networks, then the estimation process is simple and fast, but the precision and reliability of the state estimation deteriorates due to lack of uncertainty information
Solution Approach 1:
The patent replaces traditional deterministic mathematical methods with machine learning models (neural networks) that can process uncertain data. The ML models learn from historical data and provide probabilistic state estimates, substituting the need for complex physical metering infrastructure with intelligent data processing systems that handle uncertainty naturally.
Solution Approach 2:
The patent transforms the state estimation approach by changing from deterministic parameter estimation to probabilistic parameter estimation. Instead of providing single deterministic values, the system outputs probability distributions over possible states, fundamentally changing the parameter representation to include uncertainty information.
2Reliability
If probabilistic state estimation methods are used to account for uncertainty, then the reliability of state estimation improves, but the computational complexity and processing time increases
Solution Approach 1:
The patent performs preliminary training of machine learning models using extensive historical and simulated data before actual operation. This preliminary action captures complex relationships and uncertainty patterns in advance, allowing the trained models to provide rapid probabilistic estimates during real-time operation without performing complex computations at that stage.
Solution Approach 2:
The patent uses simulation data that copies real network behavior to train the machine learning models. By creating virtual copies of the power network with realistic operating conditions, the models learn from these copies and can then rapidly estimate states in the actual network without requiring time-consuming real-time simulations.
3Ease of operation
If traditional state estimation methods are used without uncertainty information, then the system operation is straightforward, but the ability to make reliable control decisions deteriorates
Solution Approach 1:
The patent adds a new dimension to state estimation by incorporating uncertainty quantification. Instead of providing only deterministic state values, the system outputs additional information about confidence levels and probability distributions, adding an uncertainty dimension that enables better decision-making while maintaining straightforward operation through automated processing.
Data Source
AI summary
The teachings herein include methods for operating a low- or medium-voltage power supply network including lines and a plurality of systems connected thereto. The method may include: providing GIS data with geographic location information relating to the power supply network; producing a branch model of the lines on the basis of the GIS data, including determining a starting probability for connections; providing system data characterizing at least one system; producing a prior model of the network by combining the branch model with the system data, including determining a starting probability for the system; generating an input state describing a possible state of the power supply network described by the prior model; carrying out a load flow simulation calculating a state variable on the basis of the input state; determining a state model of the power supply network on the basis of the state variable; and operating the power supply network based on the state model.