Taking a single bus line as the research object, this paper proposed a method to evaluate the volatility levels of passenger flows. Based on the method, the time periods and stations with high volatility level were determined. At the same time, in order to improve the performance of the bus scheduling model when the passenger flows were unstable and unpredictable, the composition of the passenger flows were described as two parts: stable passenger flow and unstable passenger flow. Based on the set pair analysis, this paper constructed a bus scheduling model considering the unstable status of passenger flow by using binary connection number. Taking no.12 bus in Harbin as an example, the model was solved by genetic algorithm in this paper. And the corresponding vehicle scheduling schemes under different deviation levels of passenger flows were presented. The results show that the bus scheduling model considering unstable passenger flow proposed in this paper could significantly save passenger waiting time and bus operation cost compared with normal bus scheduling model, especially when the fluctuation of passenger flows was more obvious. By changing the coefficient value which described the degree of instability, the model could adapt to the passenger flow at different volatility levels, and provided decision makers with scheduling options to meet different scheduling objectives.

CICTP