水力发电2026，Vol.52Issue(7)：108-113,6.
基于CEEMD-POA-BiLSTM的短期光伏功率预测
Short-term PV Power Prediction Based on CEEMD-POA-BiLSTM
摘要
Abstract
With the rapid development of renewable energy,the photovoltaic power generation,as an important component of renewable energy,its power prediction accuracy plays a significant role in the stable operation of power grids and energy management.To accurately predict photovoltaic power and solve power grid dispatching problems,a hybrid deep learning model combining Complementary Ensemble Empirical Mode Decomposition(CEEMD),Pelican Optimization Algorithm(POA),and Bidirectional Long Short-Term Memory(BiLSTM)is proposed.Initially,the Pearson correlation coefficient is used to determine irradiance and module temperature as inputs.Then,the original data is decomposed using CEEMD to extract multi-scale features,and on this basis,the POA algorithm is introduced to optimize the initial learning rate,hidden layer nodes,and regularization coefficient of the BiLSTM model.Finally,the comparative experiments are conducted on different decomposition algorithms.The results show that the proposed CEEMD-POA-BiLSTM model excels in short-term photovoltaic power prediction tasks,achieving higher prediction accuracy compared to other models.关键词
光伏功率预测/深度学习模型/互补集合经验模态分解/鹈鹕优化算法/长短期记忆神经网络Key words
photovoltaic power prediction/deep learning model/Complementary Ensemble Empirical Mode Decomposition/Pelican Optimization Algorithm/Long Short-Term Memory Neural Network分类
信息技术与安全科学引用本文复制引用
董伊浩,张杰,韩伟..基于CEEMD-POA-BiLSTM的短期光伏功率预测[J].水力发电,2026,52(7):108-113,6.基金项目
河北省创新能力提升计划项目(244C4301D) （244C4301D）
