A linear quadratic regulator with integral action of wind turbine based on aerodynamics forecasting for variable power production

Tenghui Li, Xiaolei Liu*, Zi Lin, Jin Yang, Anastasia Ioannou

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

With the increase in the share of wind energy, conventional maximum output control strategies result in difficulties in power dispatching, which motivates us to develop a novel control design for varying wind turbine (WT) power production. This study proposes a linear quadratic regulator-based (LQR) WT control to respond to power commands. Firstly, a numerical optimizer containing two algorithms and torque local linearization determines the working point and provides aerodynamic predictions for the LQR, which involves a neural network-based aerodynamics model. Secondly, a fully coupled system model based on aerodynamics forecasting can manipulate generator voltage and pitch servo input. Thirdly, the LQR combines the integral action (LQRI) to improve accuracy to minor variations. In our test of outputting the maximum, the LQRI can reduce the speed settling time by up to 25% and the output stable time by up to 28% compared with the PID-based FAST control. Regarding tracking a power reference, the proposed LQRI can achieve a steady-state error of not over 0.008 p.u. Besides, two anti-disturbance tests indicate that the LQRI can alleviate about 20% of fluctuations in the maximum capture, and varying output targets does not affect the LQRI robustness.
Original languageEnglish
Article number119605
JournalRenewable Energy
Volume223
Number of pages17
ISSN0960-1481
DOIs
Publication statusPublished - 2024

Keywords

  • Linear quadratic regulator with integral action (LQRI)
  • Model forecasting
  • Wind turbine control
  • Rotor speed control (RSC)
  • Pitch angle control (PAC)
  • Power reference point tracking (PRPT)

Fingerprint

Dive into the research topics of 'A linear quadratic regulator with integral action of wind turbine based on aerodynamics forecasting for variable power production'. Together they form a unique fingerprint.

Cite this