Volume 45: Energy Transitions toward Carbon Neutrality: Part VIII

Primary Frequency Regulation Characteristics Assessment and Allocation of Wind Farm Based on Data-driven Linear MPC Jiachen Liu, Shunqi Zeng, Zhongguan Wang, Li Guo

Abstract

This paper proposes a data-driven linear model predictive control (MPC) method to assess wind farm capability of primary frequency regulation (PFR) and reasonably allocate droop coefficient to wind turbines (WTs). The proposed method transforms the wind farm PFR nonlinear model into a linear model by using Koopman operator theory (KOT). Hence, a convex optimization problem is constructed based on a linear MPC model, which makes real-time analytical solution possible. Furthermore, the linear model coefficient matrix can be obtained by data-driven training, which is independent of complete model and accurate parameters. The case study validates that the proposed method can achieve high-accuracy assessment and allocation that the relative error is less than 1.60e-2 p.u. by only using historical operation data, and is suitable for online applications owing to the fast calculation speed, which the average assessment time is no more than 0.93s.

Keywords data-driven, droop control, Koopman, MPC, wind farm