The periodic time-varying characteristics of urban rail transit passenger flow are analyzed. Based on the traditional seasonal grouping and the monthly residual trend, the seasonal interval classification method of subway passenger flow is proposed to be applied to urban rail transit passenger flow forecasting. Establish the trend equation of the monthly passenger flow of Xi'an Metro Line 2 in 2014~2018. Calculating the monthly trend value and the monthly passenger flow residual by the trend equation, and drawing the monthly passenger flow residual curve for 2014-2018. The Fréchet distance is used to analyze the similarity between the monthly residual curves, and the K-means clustering algorithm is used to cluster the months based on the similarity. The seasonal interval based on the residual curve (SIR) is obtained. The monthly seasonal index, quarter seasonal index, and SIR seasonal index are calculated separately. The seasonal index is used to correct the passenger flow in the process of passenger flow forecasting. The corrected passenger flow is used as the forecast data to establish the Elman neural network for prediction. After the prediction, the seasonal index is applied for reverse adjustment to obtain the prediction result. The experimental results show that the improved model is smaller than the other classification methods in terms of relative error and average absolute percentage error, indicating that the proposed classification method has better performance in the short-term prediction of urban rail transit passenger flow, which provides a reference for the adjustment of the capacity of the metro transportation capacity and the formulation of the operational plan. Key words: passenger flow forecast; seasonal index; Fréchet distance; rail transit;

CICTP