Analysis and Research on Relevant Methods of Short-term Passenger Flow Forecast

Urban rail transport makes a great difference in managing urban transport construction and solving urban traffic congestion due to its characteristics of high volume, high speed and high total benefit. With the continuous development of the urban rail transport network and the continuous expansion of the network scale, with more and more metro lines, an increasingly complex network, the number of passengers also increases rapidly, which also brings with it the congestion of passenger flows in the morning and evening peak hours. Therefore, it becomes an important research that uses historical data to make a short-term prediction of transportation in a future period. As the main technology of urban rail traffic organization optimization, the accuracy of short-term passenger forecast will directly affect the suitability of the traffic organization train in the city. In this paper, three effective methods for urban train passenger estimation are introduced and summarized, which are BP topology algorithm for cloud genetic optimization, XGBoost estimation model and CNN- LSTM models. The advantages and disadvantages of these methods are also discussed. In addition, this paper analyzes the potential impact of several key objectives of short-term rail travel on passenger traffic forecasting.