基于低风速功率修正和损失函数改进的超短期风电功率预测OA北大核心CSTPCD

Wind power has strong fluctuation and randomness. In order to further improve the prediction accuracy of wind power, an ultra-short-term wind power prediction model based on power correction under low wind speed and an improved loss function is proposed. The model uses convolutional neural networks, self-attention mechanisms and bidirectional gated recurrent unit to capture long-term temporal dependencies of wind power sequences. In order to solve the problem that it is difficult for the neural network to accurately fit the waiting wind state under low wind speed, the model corrects the predicted power under low wind speed by predicting wind speed and combining wind power at the current period. To solve the stability problem of parameter training, the model introduces a multivariate nonlinear loss function to extract the correlation between sequences by improving the prediction strategy and shared weights. The results show that the proposed model is superior to the comparison model in many error indices, and can effectively improve the effect of the ultra-short-term wind power prediction.