An ultra-short-term power prediction method for wind turbines considering inflow wind speed characteristics

To address the challenges of wind speed representation distortion and insufficient dynamic response in ultra-short-term wind power prediction, this paper proposes a prediction method integrating high-fidelity inflow wind speed features with the TimeMixer model. First, considering the wind farm's terrain characteristics, turbine layout, and wake interference effects, a spatially corrected wind field is constructed to calculate the undisturbed inflow wind speed at hub height for each turbine. Second, using the incoming wind speed as the input feature, an improved TimeMixer model is developed, incorporating a dual-channel decomposition mechanism for trend and high-frequency fluctuation components. Through a multi-scale feature fusion strategy, the model's ability to jointly capture high-frequency fluctuations and low-frequency trends in turbine power evolution is enhanced. Finally, the proposed method is compared with the nacelle wind speed model, and validation is conducted using measured data from a mountainous wind farm in Qujing, Yunnan. The results show that the improved TimeMixer model with incoming wind speed input achieves improvements of 18.75%, 25%, 7.69% and 6.37% over the nacelle wind speed model in terms of root mean square error (E_rmse), mean absolute error (E_mae), r and Q_R respectively. Compared with the traditional temporal convolutional network (TCN) and long short-term memory (LSTM) network, the improved TimeMixer model exhibits superior performance in trend tracking and fluctuation response capture, with the correlation coefficient r improved by 2.46% and 5.04%, respectively. The proposed method effectively enhances ultra-short-term prediction accuracy and can provide technical support for the stable operation of power systems with a high penetration of renewable energy.