A Traction Load Forecasting Method of Electrified Railway Based on LSTM

A data-driven traction load forecasting method is proposed to better evaluate the impact of electrified railways on power quality of power grids. By analyzing the characteristics of a large number of traction load measured data, this paper combines the idea of deep learning to build a traction load forecasting model using long short-term memory (LSTM) network, including the preprocessing of input data, the construction of the deep LSTM network, the training of the model and the search of hyper-parameters. The simulation results show that the proposed method can achieve good forecasting effect, and confirm the ability of LSTM to learn the characteristics of input traction load data.