Abstract:

To determine the transport capacity by mountain scenic spots to carry tourists during holidays， an adaptation method of scenic spot transport capacity based on short-term passenger flow forecast was proposed. Based on highway traffic data， the traffic flow was converted into passenger flow， and a CNN-LSTM hybrid model for short-term passenger flow prediction was proposed； Gaussian function was used to fit the discrete data of passenger flow forecast， and breadth-first search algorithm was used to obtain the departure frequency that fitted the passenger flow curve； The paper determined the reasonable constraint conditions of vehicle operation in mountain scenic spots， considering the key parameters such as departure frequency， the number of passengers carrying capacity and traveling tiem of the trip， a capacity adaptation model by deficit function was proposed. Taking Jinsixia scenic area as a case study， the models proposed were verified. The results show that the CNN-LSTM hybrid model can effectively predict the short-term passenger flow in mountain scenic spots， and the R² of the model can reach 0.92 at the time granularity of 15 min. Compared with the traditional "full-passenger-ready-to-go" scheduling mode， the capacity adaptation model reduces the capacity demand from 57 to 28， effectively reducing the fleet supply. The research will be beneficial for short-term prediction of passenger flow in mountain scenic spots and accurate calculation of transportation capacity demand in holidays.

Key words: traffic engineering, toll collection data, adaptation of transportation capacity for scenic area, short-time passenger flow forecast, recurrent neural network, breadth-first search algorithm