Abstract: A hybrid clustering-based short-term photovoltaic (PV) power forecasting model is proposed to address the significant fluctuations and randomness in PV system power output due to weather conditions. First, Gaussian mixture model integrated with Kmeans clustering algorithm (referred to as GMKmeans) is employed to classify the original PV dataset into three distinct weather patterns: sunny, cloudy, and rainy days. Subsequently, to solve the problem of manually determining the decomposition mode number and penalty factor in variational mode Decomposition (VMD), the black-winged kite algorithm (BKA) is introduced to achieve adaptive parameter optimization for VMD. Subsequently, the optimized VMD is used to decompose the PV power time series data into multiple intrinsic mode functions (IMFs), allowing the model to deeply understand the complex patterns of PV power evolution over time. Finally, a temporal convolutional network (TCN)-bidirectional long short-term memory (BiLSTM) hybrid prediction model is constructed for each IMF component, followed by stacking and reconstructing the predicted results to achieve high-precision forecasting of the overall PV power output. Experimental results demonstrate that the proposed model significantly improves the accuracy and effectiveness of short-term photovoltaic power forecasting, offering a robust solution for renewable energy system management.