Research on bus OD prediction based on POI and passenger travel regularity

Bus OD (Origin-Destination) matrix, as the basic data of urban public transit network planning and intelligent scheduling, can provide important data support for optimizing public transportation network, improving transportation efficiency and improving urban traffic. However, due to the current one-ticket system in most cities, there is no need to swipe the card when getting off, resulting in the loss of data and the OD matrix cannot be perfected. In order to derive OD matrix quickly and efficiently, this paper presents a method combining POI and passenger travel rules. Firstly, the Point of Interesting (POI) and passenger flow data are integrated, and the POI passenger flow model is established by using neural network method. Secondly, starting from the global traffic flow, we obtain the clocking data of many bus lines in the city, and carry on statistical analysis to the data, and get the algebraic distribution of passengers between the stations. Finally, combining the algebraic distribution and POI passenger flow model, the solving OD matrix parameters is regarded as the optimal solution problem, the solving constraints are established, the line punch data and station POI are input, and the OD matrix is output. In terms of experimental results, a number of bus routes in different areas of the main urban area of Chongqing are selected, and the predicted OD matrix of each route is obtained by using the method proposed in this paper. Compared with the actual OD, the accuracy rate can reach more than 80%. Compared with the traditional travel chain and Poisson distribution, the accuracy is not only increased by more than 10%. And the time and space consumed in the calculation process are also greatly reduced, or for shorter lines, instant solutions can be achieved.