Abstract: As the installed capacity of new energy in China has historically exceeded that of thermal power, the diversified characteristics of electricity prices have become more prominent, and how to improve the accuracy of electricity price forecasting has become unprecedentedly important for electricity market participants. To this end, this paper proposes an adaptive short-term electricity price hybrid forecasting model that considers feature information. Firstly, the decomposition and reconstruction of the electricity price series are achieved by using the combined data preprocessing model of "improved variational modal decomposition-sample entropy" (IVMD-SE). Secondly, the seasonal differential autoregressive sliding average model (SARIMA) is used to predict the more obvious periodic subseries, and the whale optimization algorithm long short-term memory neural network (WOA-LSTM) is used to predict the more obvious nonlinear subseries. Finally, the predicted values of each component are summed up to get the final electricity price forecasting result. The proposed model is used to set up multiple comparative experiments based on the Australian electricity market data, and compared with the baseline model, the proposed model can achieve the lowest error index in different regions and seasons.