A composite robust adaptive control for double-fed variable-speed pumped storage units considering stochastic uncertainty

Abstract

As an advanced technology in pumped storage, the doubly-fed variable-speed pumped storage unit (DFIM-VSU) enhances flexibility, overcomes conventional constraints, and emerges as a transformative solution for power systems. However, its implementation faces challenges, including dynamic uncertainties, complex couplings, and control limitations. To address these issues, this paper proposes an adaptive control strategy to mitigate stochastic uncertainties and suppress composite disturbances, enhancing the safety of DFIM-VSU regulation. First, a coupled VSU governor-excitation model with parameter perturbations is developed. Leveraging probability density theory, a framework for dynamic response quantification is established, highlighting the regional probability distribution of key states. Building on this, a novel and efficient nonlinear disturbance observer and model reference adaptive composite controller (NDOB-MRAC) is proposed. Through state matrix reconstruction and real-time disturbance feedforward compensation, the controller can effectively optimize overshoot, maximum deviation, and regulation time in unit speed. Experimental results demonstrated that NDOB-MRAC expanded the stable domain and significantly enhanced multi-condition adaptability and dynamic performance. Notably, under low-head, high-load conditions, the proposed method reduced regulation time by 39.8 % compared to PI control while eliminating speed overshoot. These findings clarify operational uncertainties of VSUs and provide a reliable foundation for robust control in pumped storage systems.

The authors acknowledge financial support from the Smart Gird- National Science and Technology Major Project (2024ZD0801800), the National Natural Science Foundation of China (No. 52279085), the Hubei Provincial Natural Science Foundation of China (2023AFD186), the National Natural Science Foundation of China (No. U23B20143) for the research, authorship, and publication of this article. Alexandre Presas acknowledge the Serra Hunter Programme of Generalitat de Catalunya.

Peer Reviewed

Postprint (author's final draft)