AI in Grid Planning – AI Grids

In the RLI project AI Grids,  researchers are developing a tool concept to systematically summarize country-specific assumptions and technical framework conditions for energy systems and distribution grids. The processed information can then be used for electrification planning in many regions or to efficiently expand existing tools to include new regions.

Combining Data and Models

The data source is the existing literature on power grid planning in various countries. The researchers implement data extraction using a Retrieval-Augmented Generation (RAG) system, which combines queries to a Large Language Model (LLM) with information from external data sources. They first process and index the relevant body of literature. The extracted parameters are output in a structured schema and annotated with source references to enable direct further processing and validation.

Validation of the Tool Concept

To validate the concept, the researchers compare the results with the existing standard parameters of the Offgridplanner Nigeria tool. They query the original literature for the parameters used and compare the results with those obtained manually.

Transferability to Other Cases

The tool concept developed supports energy system and grid planning in regions not previously considered by efficiently analyzing relevant technical literature. The underlying methodology for structured parameter extraction can subsequently be applied to other use cases.

Project Period: May 2026 – December 2026

• Conversion of existing literature into machine-readable formats
• Data indexing
• Data retrieval using LLM
• Validation of the RAG system