Voltage support strength-constrained transmission network expansion planning for renewable-integrated grids via deep reinforcement learning

Abstract

The increasing penetration of inverter-based renewable energy sources (RES) presents significant challenges to power system stability by weakening grid strength, an issue often overlooked in traditional transmission network expansion planning (TNEP). To address this gap, this study introduces a novel planning framework that integrates data-driven uncertainty modeling with intelligent, stability-aware decision-making. We employ an improved Neural Koopman algorithm combined with the K-Means method to extract representative operational scenarios and formulate the TNEP problem as a Markov Decision Process (MDP). A Deep Reinforcement Learning (DRL) agent, specifically a Double Deep Q-Network (DDQN), is then trained to determine an optimal, sequential construction strategy. Crucially, the dynamic Short-Circuit Ratio (SCR) is incorporated as a penalty term within the agent’s reward function and enforced through an exact post hoc feasibility check to proactively ensure system stability. Validated on the high-renewable penetration standard system CEPRI-VC-88, the proposed framework yielded a plan that successfully reduced annual wind curtailment from 12.8% to 1.5% and completely eliminated scenarios with inadequate SCR at a moderate investment cost. The results demonstrate that this DRL-based approach provides a powerful and forward-looking paradigm for TNEP, effectively balancing economic, renewable accommodation, and dynamic stability objectives to foster a more resilient power grid.