广东电力2025，Vol.38Issue(1)：32-40,9.DOI:10.3969/j.issn.1007-290X.2025.01.004
基于改进蜣螂优化算法和融合注意力机制的风电功率预测
Wind Power Prediction Based on a Multi-strategy Enhanced Dung Beetle Optimization Algorithm and Integrated Attention Mechanism
摘要
Abstract
To further improve the accuracy of wind power prediction,a method is proposed that utilizes complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)to decompose the original data.This method is combined with a multi-strategy enhanced dung beetle optimization algorithm(MDBO)to optimize a wind power prediction model that integrates convolutional neural networks(CNN)and bidirectional long short-term memory(BiLSTM)networks.Firstly,the CEEMDAN decomposition algorithm is used to decompose the initial wind power data,reducing the nonlinearity and randomness of the wind power data.Then,an attention mechanism(AM)is introduced into the prediction model,where each decomposed component is predicted using the CNN-BiLSTM-AM model optimized by the MDBO algorithm.Finally,the predicted values of each component are aggregated to obtain the total prediction.Additionally,the Pearson correlation coefficient is used to calculate the correlation between environmental features and wind power,retaining strongly correlated environmental features to further enhance prediction accuracy.The proposed CEEMDAN-MDBO-CNN-BiLSTM-AM algorithm demonstrates exceptionally high prediction accuracy in wind power forecasting,with the RMSE reduced by 65.12%and 64.00%compared to the single prediction models of CNN and BiLSTM,respectively.Compared to the CNN-BiLSTM model,the RMSE and MAE are reduced by 53.20%and 53.98%,respectively,and the regression coefficient is improved by 7.581%.关键词
自适应噪声完全集合经验模态分解/风电功率预测/蜣螂优化算法/双向长短期记忆网络/卷积神经网络Key words
complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)/wind power prediction/dung beetle optimization algorithm/bidirectional long short-term memory network/convolutional neural network(CNN)分类
动力与电气工程引用本文复制引用
张旭东,汪繁荣..基于改进蜣螂优化算法和融合注意力机制的风电功率预测[J].广东电力,2025,38(1):32-40,9.基金项目
国家自然科学基金项目(61903129) （61903129）
