To enhance the accuracy and reliability of short-term photovoltaic (PV) output power forecasting, a hybrid model is proposed, which integrates principal component analysis (PCA), variational mode decomposition (VMD), and multi-verse optimizer (MVO) to optimize a support vector machine (SVM) for PV output power prediction. Initially, PCA's data analysis capabilities and VMD's data decomposition performance are leveraged to reduce the dimensionality and decompose the multidimensional training data. Subsequently, the extracted dataset is fed into an SVM prediction model optimized by the MVO algorithm to obtain PV output power forecast components for different intrinsic modes. Finally, the results of these forecast components are aggregated. The research findings indicate that the proposed model achieves mean absolute percentage errors (MAPEs) of 0.7453%, 0.5105%, and 1.0156% for sunny, partly cloudy, and rainy days, respectively. Taking partly cloudy weather as an example, the MAPE of the proposed model is reduced by 3.8207%, 2.9173%, and 1.8438% compared to the MVO-SVM, VMD-MVO-SVM, and PCA-MVO-SVM models, respectively.