PV Power Prediction Method Based on NGO-Improved VMD Decomposition and TCN^*

To address the nonlinearity and uncertainty in photovoltaic (PV) power generation data, which hinder pattern extraction and accurate prediction model development, this study proposes a TCN prediction model based on an improved Variational Mode Decomposition (VMD) method optimized by the Northern Goshawk Optimization (NGO) algorithm. First, the NGO algorithm optimizes VMD parameters to determine the optimal combination. The original complex sequence is then decomposed into intrinsic mode functions (IMFs) with simpler fluctuations. Finally, a TCN prediction model is constructed for each mode, and the individual predictions are aggregated to obtain the overall system forecast. This study provides a theoretical basis for optimizing power dispatch and improving generation efficiency.