四川轻化工大学学报（自然科学版）2025，Vol.38Issue(3)：65-74,10.DOI:10.11863/j.suse.2025.03.08
基于Yformer-DLinear的光伏功率组合预测方法
Yformer-DLinear Based Photovoltaic Power Mix Prediction Methodology
摘要
Abstract
Photovoltaic(PV)power prediction plays a crucial role in solving the problems caused by grid-connected PV.In order to improve the accuracy and robustness of PV power prediction,a combined method is constructed for PV power prediction by combining the improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN),fuzzy entropy(FE),max relevance and min redundancy(mRMR)algorithms,Yformer and DLinear.The historical PV power series are first decomposed by using ICEEMDAN and reconstructed into complex and simple components based on FE to improve the stability of the data and reduce the computational burden of the model.Then,mRMR is used to perform meteorological feature screening for each component to optimize the input features and further improve the prediction accuracy.Finally,Yformer and DLinear prediction models are established for complex and simple components respectively,and the prediction results of each model are superimposed to obtain the final prediction results.The analysis using actual PV plant data shows that the mean absolute error(MAE)of the test set is 0.653,the root mean squared error(RMSE)is 1.032,and the coefficient of determination(R2)reaches 0.958,which verifies the superiority of the proposed method.关键词
光伏功率预测/自适应噪声完备集成经验模态分解/最大相关和最小冗余算法/Yformer/DLinearKey words
photovoltaic power prediction/complete ensemble empirical mode decomposition with adaptive noise/max relevance and min redundancy algorithms/Yformer/DLinear分类
信息技术与安全科学引用本文复制引用
蔡源,吴浩,唐丹..基于Yformer-DLinear的光伏功率组合预测方法[J].四川轻化工大学学报（自然科学版）,2025,38(3):65-74,10.基金项目
四川省科技厅项目(2021YFG0313 （）
2022YFS0518 （）
2022ZHCG0035) （）
智能感知与控制四川省重点实验室项目(2019RYY01) （2019RYY01）
自贡市科技局项目(2019YYJC02 （）
2020YGJC16) （）
四川轻化工大学人才引进项目(2021RC12) （2021RC12）
