Graph Neural Network Configuration for Real-Time Grid Voltage Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Low-voltage electricity distribution networks face challenges with the integration of decentralized electricity production and new uses like EV charging, leading to voltage level increases and drops, and the deployment of smart meters is limited by communication constraints, necessitating improved network management and real-time observation.
Innovation Solution
A method using a graph neural network configured based on electrical network topology data to estimate voltage amplitudes in real-time, utilizing a graph representation to manage complex network structures, and trained neural networks to predict voltage levels across nodes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If smart meters are deployed to improve network observability, then measurement data collection capability is improved, but communication constraints limit real-time simultaneous data collection from multiple meters
Solution Approach 1:
The system performs preliminary configuration of the graph neural network with the complete network topology before operation. This allows the model to be pre-trained with structural information, enabling it to efficiently process and infer network state from limited real-time measurements without requiring simultaneous data collection from all nodes.
Solution Approach 2:
The graph neural network acts as an intermediary that processes sparse measurement data from smart meters and infers the complete network state. Instead of requiring direct measurements from all nodes, the GNN mediates between limited sensor inputs and comprehensive network observability by leveraging topological relationships to propagate information across the network.
2Adaptability or versatility
If decentralized electricity production and new uses like EV charging are integrated into LV networks, then network functionality and adaptability are improved, but voltage levels become unstable with increases and drops
Solution Approach 1:
The system implements a feedback mechanism where the graph neural network continuously estimates network state including voltage levels at all nodes based on measurements from a subset of nodes. This state information feeds back to distribution network managers who can adjust control settings on inverters, chargers, and other controllable assets to maintain voltage within acceptable ranges, thus stabilizing the network despite decentralized generation and variable loads.
Solution Approach 2:
The system changes operational parameters of network assets dynamically. By estimating voltage levels using the GNN, the system identifies nodes where voltage deviates from nominal values and adjusts parameters such as inverter reactive power output or EV charger power delivery to bring voltage back into the regulated range, thereby maintaining reliability while preserving adaptability.
3Measurement precision
If a graph neural network is configured with N layers corresponding to N nodes for real-time voltage estimation, then voltage monitoring coverage is improved, but computational complexity increases
Solution Approach 1:
The graph neural network is segmented into multiple layers, with each layer corresponding to a node in the electrical network. This segmentation allows the system to process voltage estimation in a distributed manner, where each layer handles local computations based on measurements from its corresponding node and information from connected nodes, reducing the computational burden on any single processing unit while maintaining comprehensive network coverage.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A computer-implemented method of configuring a graph neural network for managing a power grid comprising a set of nodes is provided, which comprises: obtaining topology representative data of the power grid, the topology representative data comprising a number N of nodes; configuring Ngnn neural network layers of the graph neural network, each neural network layer i comprising one or more neural networks L(k, i), and wherein the number Ngnn neural network layers corresponds to the number N of nodes of the power grid; configuring, based on the topology representative data of the power grid, neural connections of the graph neural network; and using the configured graph neural network for power grid management.