Prediction of Electricity Consumption Based on the Combination of LSTM and LassoLars

At present, with the rapid development of economy and power industry, the domestic power market has entered the stage of deepening reform. All levels of power enterprises need to forecast the power demand market. Therefore, the prediction of electricity consumption has become a very important topic. Taking the historical data of electricity consumption in Liaoning Province as an example, this paper decomposed the historical data into high-frequency data and low-frequency data by using the method of Ensemble Empirical Mode Decomposition (EEMD), introduced Long Short-Term Memory (LSTM) model to high-frequency data and LassoLars model to low-frequency data, and forecast the electricity consumption by combining the relevant Physical and meteorological factors of the power industry. Through the experiment, the combination model showed a good prediction effect.