BI Guihong, LI Yuhong, YANG Yi, et al. Ultra-short-term Photovoltaic Power Forecasting Based on DCEC-2D Clustering and MB-TSMixer-BiKSA Net Model[J]. 2025, 53(11): 99-108.
BI Guihong, LI Yuhong, YANG Yi, et al. Ultra-short-term Photovoltaic Power Forecasting Based on DCEC-2D Clustering and MB-TSMixer-BiKSA Net Model[J]. 2025, 53(11): 99-108. DOI: 10.20204/j.sp.2025.11012.
Ultra-short-term Photovoltaic Power Forecasting Based on DCEC-2D Clustering and MB-TSMixer-BiKSA Net Model
To address the issues of single clustering features, insufficient feature extraction, and low prediction accuracy in existing photovoltaic power forecasting methods, this paper proposes an ultra-short-term forecasting approach based on two-dimensional convolutional deep convolutional embedded clustering (DCEC-2D) and a multi-branch time series mixer-bidirectional gated recurrent unit-kolmogorov-arnold network-self-attention model (MB-TSMixer-BiKSA Net). By integrating DCEC-2D clustering, variational mode decomposition-empirical wavelet transform (VMD-EWT) dual-modal decomposition, and the MB-TSMixer-BiKSA Net model, the method achieves accurate clustering of typical operating conditions and effective fusion of complex features, thereby realizing high-precision power forecasting. Case study results demonstrate that the proposed method performs excellently in ultra-short-term photovoltaic power forecasting, with its mean absolute percentage error (EMAP) and coefficient of determination (R2) metrics significantly outperforming those of comparative models, exhibiting strong accuracy and robustness.
LEE Y,JEONG J.TSMixer-and transfer learning-based highly reliable prediction with short-term time series data in small-scale solar power generation systems[J].Energies,2025,18(4):765-765.
LI Y,LIU S,WU J,et al.Multi-scale Kolmogorov-Arnold Network (KAN)-based linear attention network:multi-scale feature fusion with KAN and deformable convolution for urban scene image semantic segmentation[J].Remote Sensing,2025,17(5):802- 802.