Physics-informed AI digital twins and agents for electric transmission and distribution system automation

processes. The electric grid is getting too complex and changing too quickly

for traditional utility processes. Automation is necessary for a safe,

reliable, affordable and sustainable energy system. Much like autonomous

driving for the grid, we've developed specialized physics-informed AI digital

twins to automatically and continuously "read the roads" on grid

power flow, manage it within the lanes for congestions and voltage violations,

generate routing directions in automated planning, and driving assistance with

a copilot for operations. Example use cases include automated interconnection

studies, planning automation, model validation, operational planning, real time

operational situational awareness, optimal dispatch, auto-dispatch, and event

management.

ThinkLabs has developed AI agents specialized for the

automation of mission-critical grid operations, modeled after the same design

patterns as autonomous vehicles:

• Learning system – novel physics-informed AI that learns

the complexity of grid power flow (e.g., reading the roads) and utility

workflow (e.g., learning how to drive).

• Pre-training – training the AI on numerous potential grid

scenarios (e.g., load growth, DERs, two-way power flow), such that the AI is

instantly ready for variable and unpredictable conditions in runtime.

• Speed and scale – 8760 3-phase AC unbalanced power flow

and state estimation performed in seconds to minutes, reducing entire system

studies to minutes and hours, scalable across entire networks in utility

service territories.

• Generative – AI learns from training and operational

experience and recommends optimal solutions to grid constraints, complementing

human intuition, knowledge, and trial-and-error approaches.

• Agentic – perceive, reason, and act to automate planning

and operational workflows, being trained on grid power flow, standard

practices, and operator experience.

While AI has been prevalent in grid use cases such as

forecasting, visual intelligence, and asset management, for AI to solve

fundamental challenges with grid planning and operations, it must be able to

perform power flow analysis. ThinkLabs has pioneered a method to integrate the

fields of power systems engineering with AI, known as “physics-informed AI.”

This means AI that is trained by, works with, and integrates into engineering

systems (e.g., CYME, PSSE, PSLF, ADMS).