Hierarchical power control of a large-scale wind farm by using a data-driven optimization method.

Area of Science:

• Electrical Engineering
• Renewable Energy Systems
• Control Systems

Background:

• Large-scale wind farms require efficient distribution of automatic generation control (AGC) signals to numerous wind turbines (WTs).
• The wake effect among WTs complicates the computation of high-quality dispatch schemes.
• Existing methods face computational expense in optimizing real-time AGC signal distribution.

Purpose of the Study:

• To develop a novel hierarchical power control (HPC) strategy for large wind farms.
• To reduce the computational burden associated with AGC signal distribution and WT dispatch.
• To improve the quality of dispatch schemes considering wake effects.

Main Methods:

• A hierarchical power control (HPC) framework is proposed, leveraging WT geographical layout and electrical connections.
• The real-time AGC signal is distributed to decoupled groups based on regulation capacities.
• Data-driven surrogate-assisted optimization and dynamic surrogate-based local search are employed for WT-level signal distribution.

Main Results:

• The HPC method significantly reduces computation time by minimizing objective function evaluations.
• High-quality dispatch schemes are achieved through efficient local search mechanisms.
• The technique's effectiveness is validated across various AGC signals, wind speeds, and directions.

Conclusions:

• The proposed hierarchical power control (HPC) offers an efficient and effective solution for AGC in large wind farms.
• This approach mitigates computational challenges posed by complex wind farm dynamics and wake effects.
• The method ensures high-quality dispatch schemes while optimizing control signal distribution.