Renewables
AI framework optimizes grid expansion for high renewable penetration
Researchers have developed a deep reinforcement learning approach to manage grid strength and lower renewable curtailment during transmission planning.
Key points
- A new transmission expansion framework treats dynamic grid strength as an actively managed indicator.
- The method employs a Koopman operator framework and a double deep Q-network agent.
- Testing on a standard 100-bus network cut annual wind curtailment from 12.8 percent to 1.5 percent.
What happened
Engineers at an energy research institute have introduced a novel transmission network expansion planning approach designed to maintain stability as solar and wind assets displace conventional generation. Inverter-based resources reduce rotational inertia and short-circuit capacity, weakening voltage support during disturbances. To address this mathematical and operational challenge, researchers utilized a deep reinforcement learning agent to build new lines sequentially while monitoring dynamic grid strength.
Why it matters
Traditional planning models struggle with the non-convex relationship between network topology, alternating-current power flow, and stability indexes like the short-circuit ratio. By employing a neural Koopman embedding to process operational data and graph neural network surrogates to estimate operational costs and stability margins, the framework bypasses the computational limits of classical programming. When tested on a 100-bus standard test network, the double deep Q-network agent successfully lowered wind curtailment and achieved lower total planning costs than baseline optimization methods.
What comes next
The strategy enforces dynamic stability requirements through both soft constraints during policy search and exact feasibility verification for every node. As grids worldwide incorporate higher shares of variable generation, such AI-driven tools offer a pathway to maintain voltage stability without compromising cost-efficiency.
Original reporting
This article draws on the following reports. Read them in full on the publishers’ sites.
Context from CEII
Renewables are the fastest-growing part of the world’s energy system. The COP28 goal is to triple capacity by 2030, and solar mini-grids are the fastest route to power for millions of people without electricity.
27.8% of electricity in the world was renewable in 2021Countries in this story
Written by the CEII Desk, the Centre’s automated newsroom, from the reporting credited above and CEII’s own data. Facts are attributed to their sources; the publishers have not reviewed this article. Spotted an error? Tell our editors.
