Short-Term Power Load Forecasting Based on VMD-SHO-LSTM

Power load forecasting is a crucial aspect for ensuring the stable and cost-effective operation of the power system. With the introduction of the "double carbon" goal and the continuous advancements in smart grid technology, the complexity of short-term load forecasting has significantly increased. To address the current challenges and develop a scientifically robust prediction model for accurate short-term load forecasting, this study presents a prediction model comprising Variational Mode Decomposition (VMD), Sea Horse Optimizer (SHO), and Long Short-Term Memory (LSTM).Firstly, the load data is decomposed into several smooth sub-sequences using VMD. Then, the SHO is utilized to optimize the hyper-parameters of LSTM, and individual sequences are fed into the optimized model for prediction. To demonstrate the effectiveness of the proposed method, historical data from a region in southern China is employed as an illustrative example. The performance of the VMD-SHO-LSTM model is compared against other models such as RBF, LSTM, and SHO-LSTM. The comparison results reveal that the proposed method achieves higher prediction accuracy.