Abstract: With the growing penetration ratio of new energy, in this paper we propose a combined aggregation prediction model based on wavelet decomposition(WD), crested porcupine optimizer(CPO) and variational mode decomposition (VMD) to address the issue of more fluctuating waveform and the difficulty in predicting net load compared to traditional load. The model comprises a quadratic decomposition module and a neural network prediction module. Since the net load exhibits strong volatility, firstly, the original data is decomposed into high frequency components and low frequency components by wavelet decomposition combined with permutation entropy. The high frequency components are further decomposed using the variational mode decomposition method improved by the crown porcupine optimization algorithm. Secondly, the attention mechanism is integrated into the long short-term memory network to enhance its ability to focus on critical information. Finally, the high and low frequency components are integrated into the improved neural network through matrix aggregation and vector aggregation to obtain final prediction results. The example analysis demonstrates that the proposed model achieves better prediction effect compared to traditional prediction models, exhibiting an enhanced effect at trend turning point and better handling of sudden change in net load.