Forecasting method of change trend of single-line bus operation state based on multi-source data

This paper proposes a spatio-temporal processing and parsing method for multi-source heterogeneous data and clarifies the definition and characteristics of bus operation states. It selects indicators to characterize the changing trend of bus operation states and completes the prediction of the changing trend of bus operation states for a single line by predicting the bus vehicle arrival time and operation state characterization index. The accuracy of the BP-DS model prediction is compared with both the traditional statistical theory model and the BP neural network model. By comparing the prediction results of the three models, the effectiveness of the BP-DS model can be evaluated and validated. Additionally, based on the model prediction results, further suggestions for improving the bus operation states are proposed. The research results have important theoretical and practical significance in enriching and enhancing existing theories and developing more effective improvement strategies for bus operation states, as well as decision support for the intelligent bus management urban development.