基于IWOA-ITransformer-LSTM的风电功率短期预测

Short-Term Wind Power Prediction Based on IWOA-ITransformer-LSTM

中文关键词:风电预测  改进的鲸鱼优化算法  长短时记忆  改进的Transformer

keywords:wind power forecasting  IWOA  LSTM  Itransformer

中文摘要:风电功率数据具有高度的随机性和非线性特征，给预测模型的构建带来了巨大的挑战。为此，提出了一种基于改进的鲸鱼优化算法（improved whale optimization algorithm，IWOA）、改进的Transformer（improved transformer，ITransformer）和长短时记忆（long short-term memory，LSTM）神经网络的混合模型，以实现对风电功率的高精度短期预测。首先，IWOA用于优化ITransformer模型的超参数，增强了模型的性能；接着，通过ITransformer模块自注意力机制有效地捕捉了风电功率数据中的长时依赖性特征，并将这些特征传递给LSTM模型以进一步捕捉数据中的短期动态变化；最后，仿真结果验证了所提出方法的有效性和优越性。

Abstract:Wind power data exhibits highly random and nonlinear characteristics，which pose great challenges to the construction of prediction models. To address this problem，a hybrid model based on the Improved whale optimization algorithm （IWOA），improved transformer （ITransformer） and long short-term memory （LSTM） neural network is proposed to achieve high-precision short-term wind power prediction. First，IWOA is used to optimize the hyperparameters of the ITransformer model，which enhances the model performance. Second，the ITransformer module effectively captures the long-term dependency features in wind power data through the self-attention mechanism，and transmits these features to the LSTM model to further capture the short-term dynamic changes in the data. Finally，simulation results verify the effectiveness and superiority of the proposed method.