Power System Load Forecasting Method Based on LSTM Network

Abstract

Abstract Scientific load forecasting methods and accurate forecasting results are an important basis for the power system planning department work, as well as the basis and guarantee for correct decision-making on investment and construction. This paper has thoroughly studied the basic principles of recurrent neural networks and their shortcomings that they cannot solve long-term dependence problems, and a power system load forecasting method based on LSTM networks is proposed. Through the built-in two memory state lines of long term and short term, the problems that it is difficult to capture the long-term development trend and short-term fluctuation characteristics of data in time series prediction were solved. The results of case analysis show that this method can effectively improve the accuracy of power system load prediction.