Xu Y.-NHuang Y.-CChen Y.-WChuang C.-HYUAN-YIH HSU2023-06-092023-06-09202217521416https://www.scopus.com/inward/record.uri?eid=2-s2.0-85128193218&doi=10.1049%2frpg2.12472&partnerID=40&md5=da0d63c6501de81ca1efe517a9f89367https://scholars.lib.ntu.edu.tw/handle/123456789/632253This paper proposes a generalized piecewise linear time-variable droop frequency controller with a ramp-up period for a de-loaded wind turbine generator. The authors present an analytical model using a simplified system frequency response (SFR) to explain the need for a ramp-up period under the conditions of variable wind speed. In order to achieve the highest frequency nadir under different operating conditions, the durations and peak value of the time-variable droop function must be adapted in real time. This adaptation is achieved using an artificial neural network (ANN) based on current system variables such as wind speed, penetration, pitch angle, wind speed change, and outage capacity. Dynamic frequency responses for a power system subject to generator outages and wind speed changes are simulated using MATLAB/Simulink to demonstrate the effectiveness of the proposed droop controller. These results show that the frequency nadir can be improved by the proposed algorithm through adaption of the durations and peak value. © 2022 The Authors. IET Renewable Power Generation published by John Wiley & Sons Ltd on behalf of The Institution of Engineering and Technology.[SDGs]SDG7Controllers; Frequency response; Neural networks; Outages; Piecewise linear techniques; Turbogenerators; Wind turbines; Frequency controllers; Frequency nadirs; Linear time; Piecewise linear; Piecewise-linear; Ramp up; Time variable; Up period; Wind speed; Wind turbine generators; Windjournal article10.1049/rpg2.124722-s2.0-85128193218