Coordinated power smoothing control for wind storage integrated system with physics-informed deep reinforcement learning

Abstract

The Wind Storage Integrated System with Power Smoothing Control (PSC) has emerged as a promising solution for efficient and reliable wind energy generation. However, existing PSC strategies exhibit several limitations. Many fail to capture the cooperative interactions and distinct control frequencies between wind turbines and battery energy storage systems (BESSs). In addition, the impacts of wake effects and battery degradation costs are often overlooked. This paper proposes a novel multi-agent coordinated control framework to address these challenges, which explicitly integrates a wake model and a battery degradation model to construct a more realistic operating environment. The problem is formulated as a multi-agent Markov decision process (MMDP), where the wind farm and the BESS agents pursue complementary objectives to achieve optimal control. Furthermore, a Physics-informed Neural Network-assisted Multi-agent Deep Deterministic Policy Gradient (PAMA-DDPG) algorithm is introduced, embedding a partial differential equation of power fluctuation as a physics-guided loss term to accelerate learning and enhance physical consistency. Simulations using WindFarmSimulator (WFSim) in four scenarios demonstrate that the proposed method outperforms benchmark approaches, achieving an 11% increase in total profit and a 19% reduction in power fluctuation. These results effectively address the dual objectives of economic efficiency and grid reliability.