To improve the accuracy of the ultra-short term wind power prediction
an ultra-short-term wind power prediction model based on a Long Short-Term Memory (LSTM) neural network algorithm combined with the Transformer (TRF) model is proposed. The cosine annealing with warm restarts (CAWR) strategy is introduced to optimize the proposed prediction model to prevent the prediction model from converging to local optima. Firstly
the density-based spatial clustering of applications with noise (DBSCAN) and random forest (RF) methods are employed for anomaly detection and data imputation. Secondly
the CAWR-LSTM-TRF combined model is utilized to extract wind power features. Finally
the experiments for ultra-short-term wind power forecasting are conducted. The results of the study show that the symmetric mean absolute percentage error of the combined LSTM-TRF prediction model optimized by CAWR is reduced by an average of 0.46 percentage points compared to the LSTM-TRF model. Therefore
the proposed model achieves higher prediction accuracy and superior forecasting performance.
ESTER M, KRIEGEL H, SANDER J, et al.A density-based algorithm for discovering clusters in large spatial databases with noise[C]//Proceedings of the 2nd International Conference on Knowledge Discovery and Data Mining Protland, 1996: 226-231.
VASWANI A, SHAZEER N, PARMAR N, et al.Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook: Curran Associates Inc, 2017: 6000-6010.