Demand side load prediction method and system, computer equipment and storage medium

The invention discloses a demand side load prediction method and system, computer equipment and a storage medium, and relates to the technical field of power load prediction, and the method comprises the steps: obtaining historical demand side load data, and making a sample data set; performing stationarity analysis on the sample data set by using ADF test, decomposing an unstable data feature set by using STL, and creating an expanded feature data set; according to a pre-created LGBMRegressor model, the expanded feature data set is screened, and an optimal feature combination is selected; and inputting the obtained optimal feature combination into a DeepAR model to obtain a prediction result of the demand side load power, and carrying out evaluation analysis. According to the method, errors in model training are reduced, the credibility and stability of the model and the prediction accuracy of the demand side load power are improved, a more reliable reference basis is provided for actual operation, and optim