A novel combination model for ultra-short-term wind speed prediction

Abstract Accurate and fast ultra-short-term wind speed prediction is the key to wind farm operation control.A model of wind speed prediction, utilizing Variational Mode Decomposition (VMD), Sparrow Search Algorithm (SSA), and Long Short-Term Memory (LSTM), is proposed to tackle the issue of inadequate precision in current ultra-short-term wind speed forecasts. In response to the problem of complex wind speed datasets and difficulty in selecting VMD parameters, the archimedean optimization algorithm (AOA) is used to optimize the modal component values and penalty factor, achieving adaptive selection of VMD parameters. Furthermore, in response to the problem of optimizing LSTM parameters in wind speed prediction, the SSA algorithm is used to optimize its parameters, achieving the goal of optimizing hidden layer neural units, iteration times, and initial learning rate parameters.The VMD-SSA-LSTM model, combining VMD, SSA and LSTM, has been employed to accurately forecast ultra-short-term wind speed in wind farms based on past wind speed data. Experiments have demonstrated that this model has improved the accuracy of wind speed prediction to around 97%, thus providing a novel approach to ultra-short-term wind speed forecasting in wind farms.