Electric Vehicle Scheduling Based On Stochastic Traffic Conditions

As the process of urbanization and transportation motorization continues to accelerate,the problem of urban pollution has become increasingly prominent.At this juncture,many countries and cities at home and abroad are vigorously promoting the development of electric vehicles to promote urban energy conservation and emission reduction,especially in the field of public transportation.In recent years,the proportion of electric buses in China’s urban public transportation has continued to rise,and the proportion of pure electric bus operating lines has continued to rise.However,electric buses have brought new challenges to bus operators: how to prepare a scientific and reasonable vehicle scheduling plan for electric buses? A scientific and reasonable vehicle dispatching plan will not only help reduce the operating costs of public transport companies,but also improve public transport operations and service levels.However,compared with traditional fuel buses,electric buses have insufficient cruising range,long charging time,and limited construction of supporting facilities related to supplementary energy.Therefore,it is necessary for us to carry out research on the electric vehicle scheduling problem(EVSP)based on the characteristics of electric vehicle.This paper designs a gainful network flow model with uncertainty to better represent the electric vehicle scheduling problem.The compatibility between trips is subdivided into time feasibility and energy feasibility.And the time feasibility is redefined according to trip time distributions instead of deterministic trip time values as traditionally done.In order to prevent electric buses from running out of power during operation,and to minimize operating costs as much as possible while ensuring on-time performance,we have established a mathematical optimization model with the primary goal of minimizing the size of the fleet and the secondary goal of minimizing the cost of operating vehicles while ensuring on-time performance.At the same time,this paper designs a large neighborhood search algorithm that hybridizes with the simulated annealing algorithm(LNS-SA)to solve the proposed model.According to the characteristics of the EVSP,three remove operators and three insert operators are designed.Later,on the basis of LNS-SA algorithm,an improved adaptive large-scale neighborhood search algorithm(IALNS)was designed.In IALNS algorithm,the weights of the operators will be updated periodically according to their performance in iterations.At the end,based on real cases in City X,the feasibility and effectiveness of the algorithm and model were verified,and some sensitivity analysis were carried out.Experimental results showed that both the LNS-SA and the IALNS algorithms can solve EVSP effectively under reasonable on-time performance,and the IALNS algorithm achieved better solving performance by introducing the operator adaptive mechanism.Moreover,the stochastic model can better describe real-world problems than the traditional deterministic model.