How can the low voltage (LV) grid topology be calculated using an AI? The connectivity of each meter with the corresponding transformer station, panel, line, and phase must be identified based on the recorded currents from the meters.
Objectives
Automatic calculation of LV grid topology, determining meter connectivity with close to 100% reliability. Automatic updates implemented in case of topology changes, ensuring precise customer-to-line-phase connection point associations. This enhancement streamlines DSOs’ digital asset inventory management.
Use case
By analysing data from smart meters and supervisors at Transformer Substations, the algorithm calculates the electrical connection of each meter in a short period of time. Furthermore, this approach remains agnostic to the type of meters and communication technologies employed.
Infrastructure
On premise Cloud
Technology
Machine learning and deep learning
Data
High-resolution time series of electrical data from meters and Transformer Substations.
Resources
Multidisciplinary team of experts in telecommunications, data scientists and professionals from the electrical sector. Framework of servers, storage systems and data management tools. Effective cybersecurity and data anonymisation measures to ensure integrity and regulatory compliance.
Difficulties and learning
The challenges we faced taught us the importance of setting realistic goals and managing time effectively. In addition, we learned to distinguish between an algorithm and artificial intelligence as a broader structure, with a solid infrastructure and an independent value chain.
KPIs (business impact and metrics of the model)
The topology calculation is currently unattended and is executed daily on more than one million meters with a success rate of over 99.95%.
Funding
Mixed financing of public grants and private investments.