35 Study on Incident Detection Technology for Transit Line Operation Based on GPS Available to Purchase

An incident detection technology is studied for public transit line operation in this paper. Through the analysis of types and selection principles from current detection parameters, the bus speed is proposed to be the detection parameter for transit operation incident, by analyzing the operation characteristics of urban public transportation. Firstly, a deviation analysis method is adopted in the detection algorithm for abnormal situation in transit operation, by analyzing the existing traffic incident detection algorithms' advantage and disadvantage. Secondly, a short-term bus speed forecasting model is established based on Radial Basis Function Neural Network. An incident detection model for transit line operation is built, by comparing the predicted bus speeds and the collected speeds from GPS in the bus. Real data from one bus line in a city are applied to test the technology's availability. Good agreement from predicted and observed data is found, thus the results prove that the technology is feasible.