Long Term Load Forecasting Method for Holiday Correction Based on LSTM

Pre year monthly load forecasting is the basic data for the power system to formulate medium and long-term power generation plans, and its prediction accuracy directly affects the rationality of power generation plan formulation and other work. However, public holiday factors, as an important factor affecting load accuracy, are not adequately considered by existing pre year monthly load forecasting algorithms, resulting in large load forecasting errors for the months where public holidays occur, and historical load data obtained from power grid companies is insufficient to meet the needs of load forecasting accuracy. Therefore, it is necessary to preprocess the raw data and carefully consider public holiday factors. This article takes the power grid in a certain region as the object, processes the raw data to improve data quality, and proposes a prediction method based on the influence factors of normal values according to the obtained public holidays load fluctuation characteristics. The numerical simulation and result analysis show that after data processing and public holiday factors correction, the load forecasting error can be effectively reduced.