The impact of wind evolution and filter design on lidar-assisted wind turbine control
| Main Authors: | Feng Guo, David Schlipf, Yiyin Chen |
|---|---|
| Format: | info Proceeding Journal |
| Bahasa: | eng |
| Terbitan: |
, 2021
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/4985412 |
Daftar Isi:
- Presentation at the Wind Energy Science Conference 2021 Lidar-assisted wind turbine control has been proven to be beneficial for wind turbines. This technology uses the preview of rotor-effective wind speed obtained from lidar measurements in front of the wind turbine. With the previewed wind speeds, additional feed-forward loop is added to the conventional feed-back loop, which allows the turbine to pre-react to the disturbance [1]. However, due to limited upstream measuring positions and the fact that wind turbulence evolves continuously along the longitudinal direction, usually only the lower frequency components of the lidar measurement can be used for turbine control. The quality of the wind preview can be indicated by how much of the usable frequency components for control [2], and more specifically the coherence structure of the turbulence. In different atmospheric conditions, the strength of wind evolution could change [3], resulting in different coherence between lidar estimated and the actual rotor effective wind speed. However, research regarding how the wind evolution conditions influence the feed-forward control performance and filter design is still rare. Previously, the wind preview quality of lidar during short field-testing campaigns has been identified offline prior to using the lidar signal for control. The filter is designed with a frequency-based correlation study by comparing the rotor-effective wind speed estimated from turbine data to the one provided by the lidar [4]. An online application of the frequency-based correlation study however is hard to implement due to the uncertainties of estimating rotor-effective wind speed from turbine data, the amount of data necessary, and the sensitivity of the frequency-based method itself. A new statistic-based method to estimate the wind preview quality by only using the data from a pulsed lidar system without the need of wind turbine data has been developed [5]. Based on the simulated line-of-sight wind speeds from lidar, a statistic model is firstly used to detect the correlation decay in the turbulence field. Then the coherence decay constants can be derived which could be further used to estimate the transfer function between the lidar estimated and the actual rotor-effective wind speeds [1]. In order to test this algorithm, we define different wind evolution conditions based on the observation by [2], and several realizations of the wind field and lidar measurements simulations are performed. Figure 1. shows that this method is able to accurately distinguish the transfer function, where the mean estimations of different realizations are close to the defined analytical transfer functions and the confidence level is promising. With the estimated transfer function, an optimal Wiener filter could be designed that can minimize the measurement error [2] [6], i.e., the error between lidar estimated and actual rotor effective wind speed. As shown by Figure 2, if a non-optimal filter that under or overestimate the evolution condition is applied, the performance of feed-forward control can be contaminated. More importantly, the resulting variance in fore-aft tower bending moment with feed-forward control can be higher than the conventional feed-back-only control when a wrong filter is applied in strong evolution conditions. Here the longitudinal coherence decay constant “a” is used to represent the strength of evolution. In this study, we aim at investigating the impact of wind evolutions on lidar-assisted feed-forward control and proposing suggestions regarding robust filter design under various evolution scenarios. References for the abstract [1] Schlipf, D., 2016. Lidar-assisted control concepts for wind turbines. Dissertation. [2] Simley, E. and Pao, L., 2013, June. Reducing lidar wind speed measurement error with optimal filtering. In 2013 American Control Conference (pp. 621-627). IEEE. [3] Chen, Y., Schlipf, D. and Cheng, P.W., 2020. Parameterization of Wind Evolution using Lidar. Wind Energy Science Discussions, pp.1-35. [4] Schlipf, D., Fleming, P., Haizmann, F., Scholbrock, A., Hofsäß, M., Wright, A. and Cheng, P.W., 2014, December. Field testing of feedforward collective pitch control on the CART2 using a nacelle-based lidar scanner. In Journal of Physics: Conference Series (Vol. 555, No. 1, p. 012090). IOP Publishing. [5] Guo, F., Schlipf, D. 2021. Lidar wind preview quality estimation for wind turbine control. Submitted to 2021 American Control Conference [6] Vidyasagar, M., 2017. Signals, Systems & Inference [Bookshelf]. IEEE Control Systems Magazine, 37(4), pp.97-98.