Passenger Flow Prediction of Urban Rail Transit Stations Based on EMD-SSA-LSTM Model
Based on EMD and SSA algorithms,the LSTM neural network was optimized and a new combined forecasting model was proposed.EMD algorithm was used to reduce the interference of data noise,and the short-term passenger flow data was decomposed into multiple IMF and a residual.The SSA algorithm was adopted to optimize the number of hidden layer neurons,learning rate and itera-tion times of LSTM network.The optimized LSTM model was used to predict each IMF,and the final prediction value was obtained by summing the prediction results of each IMF.The passenger flow data of East Railway Station,the station with the largest passenger flow in Hangzhou,was used to verify the results,and compared with the prediction results of BP neural network,LSTM neural network and SSA-LSTM model.The results show that the forecasting errors of EMD-SSA-LSTM combined model are lower than the other three models,and the determinable coefficients between the predicted value and the real value of working days and non-working days are 0.9995 and 0.998 respectively,which verifies the effectiveness of the combined model proposed in this paper and improves the fore-casting accuracy.
short-term passenger flow forecastEMD and SSA algorithmsLSTM neural networkcombined prediction model
万方数据
维普
