Optimization Method Of Bus Scheduling Based On GPS Data

As the core of bus operation management,bus scheduling determines the operation cost,work efficiency and service level of the bus system to a certain extent.Nowadays,the wide application of intelligent technology,such as GPS,AVL and GIS,provides more accurate and comprehensive data support for bus scheduling.However,the traditional method has some limitations and bottlenecks in the application,processing and feedback of information.Therefore,it is necessary to carry out the related research of intelligent bus operation and scheduling methods,so as to improve the operation efficiency and service quality of bus system,promoting the construction and development of public transport.In this paper,it takes urban bus as the research object.It establishes an optimization model of bus departure interval considering random travel time,and proposes an improved dynamic bus holding strategy.The contents and conclusions are as follows:Firstly,the collection principle and processing method of GPS data are introduced.Based on the preprocessing of initial data,two data mining methods,cluster analysis and distribution fitting are applied to analyze the operation characteristics of bus system.The results show that the travel time of the bus is random and fluctuate.The log-normal distribution model can better fit the distribution of bus travel time data in different time periods.Secondly,in view of the lack of random consideration of bus travel time in current research,the optimization model of bus departure interval considering random travel time is established based on the weighted minimum value of passenger and bus company's cost expectation and its average deviation value.The SA-PSO algorithm and the Monte Carlo simulation method are used for solving the model.The case study shows that the optimized model has stronger anti-interference ability,and it can be better close to the actual operation situation and improve the reliability of the bus service,which proves that the model is reasonable and feasible.Finally,take the bus holding strategy as the research focus and combining LS-SVM algorithm,a dynamic holding strategy based on short-term travel time prediction is proposed.The strategy takes the vehicle's punctuality at the current stop and the next stop into account at the same time,and then the judgment is made according to the holding condition and the corresponding holding time is calculated.In particular,the capacity and the arrival rate of the stop is considered as the main factor of the holding condition,so as to reduce the impact on other traffic flows.Simulation experiments are carried out to verify the effectiveness and analyse sensitivity of parameters.The results show that the strategy can efifectively improve the bus operation uniformity and reduce the average waiting time of passengers.It has significant effect on preventing the phenomenon of bus bunching and large interval and improving the reliability of bus.It may provide a scientific and reasonable decision for bus scheduling and passenger.