British consortium to collaborate on Neural BB for EMT simulation
Image courtesy 123rf
The UK’s National Energy System Operator (NESO), consultant Transmission Excellence, The University of Bristol and The University of Bath are working together on Neural BB, to a methodology that can use neural networks to create a trained model for Electromagnetic Transient (EMT) simulation.
According to NESO in a release, EMT models are more detailed, complex and require more processing power and time to run than RMS (Root Mean Square) models.
Hence, the company is collaborating on the methodology, known as Neural BB, which will be able to use neural networks to create a trained model for EMT, which is faster and provides the same accuracy of the complex parent model due to machine learning.
Have you read:
NESO introduces tool for data-secure EMT analysis
GB’s NESO leads project on network security in a quantum future
An influx of IBR
According to the company, the model comes as the diversity of generator types providing power in the UK is set to increase. The most common of these will be Inverter-Based Resources (IBR), generators connected to a converter rather than a spinning turbine that is synchronous to the grid such as battery energy systems, wind and solar generators.
IBR have faster dynamic behaviours than synchronous generators such as gas, nuclear or biomass. The most common analysis methods NESO currently use are Root Mean Square (RMS) based, which they say excel at modelling synchronous generators and a lot of system events accurately and quickly. However, this may not be suitable for faster transient behaviour, when the power supplied changes momentarily, which is common in a network dominated by IBRs.
EMT simulation can capture the faster fluctuations in power but takes longer to run simulations and more complex models.
This is where Neural BB comes in, bridging the gap using lightweight models that offer the same level of accuracy with lower simulation times.
By utilising this method of machine learning, NESO says they can create accurate models that can be used in future stability studies. Additionally, accurate EMT simulation could reduce costs by increasing visibility of potential issues, accurately representing different generation types, reducing complexity and allowing more detailed models to be used more often.
Successful delivery of the project will allow margins of safety to reduce, which could lead to additional savings, particularly across system boundaries that might be constrained on the network.
