How a Power Utility achieved 95% Consumer-to-DT mapping accuracy with Applied AI

At a Glance

Key challenges faced by the utility:

• Consumer-to-DT connections were not properly established: The utility had consumer meter IDs, location coordinates and outage data, but lacked reliable mapping of which transformer supplied each consumer.
• Location alone could not identify the connected DT: Consumer and DT coordinates showed nearby transformers but could not confirm the actual connection among several possible DTs.

Grid in Action (How It Was Implemented)

1. Location-based DT identification

Applied AI used consumer and transformer coordinates to shortlist up to 10 potential DTs within a 1 km radius of each consumer.

2. Outage-based mapping validation

Applied AI was used to compare consumer and DT outage start and restoration times to identify the most likely connection.

3. Mapping status and exception identification

Each consumer received a probable DT assignment or an unresolved status with the reason recorded when sufficient evidence was unavailable.

4. Dynamic consumer-to-DT mapping

The utility could rerun indexing with updated consumer, DT and outage data as network connections changed, with revised mappings available for MDM, CIS and GIS.

5. Consumer-to-feeder mapping

The mapping extended from consumers to DTs and feeders, supporting energy accounting and loss analysis further upstream.

Grid in Numbers

• ~7 lakh smart meters covered by the utility

• 42% of existing consumer-to-DT mappings found incorrect

• ~2.94 lakh incorrect mappings identified across the smart meter base

• 58% → 95% consumer-to-DT mapping accuracy after Applied AI implementation

• +37 percentage points improvement in mapping accuracy