ChatGrid: Power Grid Visualization Empowered by Large Language Model

Abstract

This paper presents a novel open framework, ChatGrid, for easy, intuitive, and interactive geospatial visualization of large-scale transmission networks. ChatGrid uses state-of-the-art techniques for geospatial visualization of large networks including 2.5D views, animated flows, hierarchical and level-based filtering, and aggregation to provide visual information in an easy, cognitive manner. The highlight of ChatGrid is a natural language query based interface powered by large language model (ChatGPT) that offers an enhanced interactive experience whereby the user asks a question and ChatGrid provides the information both in text and visual. We discuss the architecture, implementation, design decisions, and usage of large language model for ChatGrid.