Research on multi-time scale power decomposition and predictive control methods for hybrid flywheel energy storage systems

• A multi-time scale hybrid FESS configuration strategy was designed, where energy-type FESSs were allocated to handle low-frequency power fluctuations with large energy demands, and power-type FESSs were deployed for high-frequency fluctuations requiring fast response. This ensured both effective smoothing and dynamic balance between output capability and response time. • A DC-side grid-connection model with parallel topology and master–slave control structure was established for the hybrid FAESS. The system adopted layered control: the upper layer implemented MPC-based optimization of charge/discharge scheduling using extracted high/low frequency components, while the lower layer used a dual-loop control strategy for accurate regulation of flywheel speed and output power. • The proposed SGMD decomposition method enabled precise separation of frequency components in wind power, which allowed for more targeted control in the hybrid FESS. This improved the efficiency of power allocation and reduced the frequency and depth of unnecessary charge/discharge cycles. • Simulation results show that the proposed strategy significantly improved wind power smoothing performance, reduced the fluctuation range of FESS SOC, and enhanced overall energy storage utilization. Specifically, Strategy II showed superior performance compared to Strategy I in terms of power balance, SOC stability, and loss reduction. The large-scale integration of wind power introduces significant challenges to the active power balance in power systems. Deploying energy storage systems (ESS) at wind farm grid connection points is an effective solution to mitigate output fluctuations and improve the penetration of renewable energy. Among various ESS technologies, flywheel energy storage (FES) has attracted attention for its fast dynamic response, high reliability, and environmental friendliness, making it highly suitable for wind power applications. However, conventional single-mode flywheel systems often fall short in addressing multi-time-scale regulation requirements and broad power variation scenarios. To overcome this limitation, this paper proposes a hybrid flywheel energy storage array (HFESA) scheme capable of operating across multiple time scales ranging from milliseconds to several minutes. The system integrates a 1 MW/10 MJ short-duration power-type flywheel and a 0.3 MW/100 MJ long-duration energy-type flywheel to accommodate different frequency components of wind power output. To facilitate optimal time-scale decomposition, Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Symplectic Geometry Mode Decomposition (SGMD) are employed to extract high- and low-frequency components, which are subsequently allocated to the corresponding flywheel units. A two-dimensional micro-element kinematic model is developed for the power-type flywheel, and a Model Predictive Control (MPC) strategy is adopted to realize dynamic power dispatch. For the minute-scale energy-type flywheel storage array, a state-space model is established, and the power allocation is implemented using a model predictive control (MPC) algorithm.Simulation results based on real wind farm data demonstrate that the proposed HFESA control scheme significantly improves output smoothness, minimizes unnecessary charging/discharging cycles, and enhances the overall energy utilization efficiency. The system effectively balances flywheel output limits and reduces power fluctuations during grid integration, demonstrating its potential for improving the stability and efficiency of wind power systems.