 This paper proposes a novel hybrid algorithm combining similar days selection, SDS, Empirical Mode Decomposition, EMD, Long Short-Term Memory, LSTM, Neural Networks, and Extreme Gradient Boosting, XGBoost, to construct a load forecasting model. The XGBoost algorithm was used to select the most similar days in the past, while EMD decomposed the data into intrinsic mode functions, IMFs. Finally, separate LSTM neural networks were trained on each IMF and residual to predict the load. The results show that this approach outperforms other methods in terms of accuracy and precision. This article was authored by Huiding Zheng, Jebin Yuan, and Long Chen.