| Taxipooling can improve the efficiency of taxi, so it can ease the difficulty of taking taxi in the large and medium-sized cities in China without increasing the taxi quantity. However, there is no scientific model to guarantee the rationality of the route and protect the benefit of all the aspects in taxipooling. In recent years, taxi reserve service comes into the vision of people, and gradually cultivates people’s habit of making an appointment to take a taxi. Taxi reserve pattern has the congenital superiority in user requirements gathering, how to optimize the route of taxipooling based on the users’ travel demands under the background of the taxi reserve, to improve the utilization rate of vehicles, ease road congestion and reduce environmental pollution has certain practical significance.In this paper, on the national natural science fund project, establish the model of taxipooling routing optimization, and design the improved genetic algorithm to solve the model. Solve the model with matlab and analyze the model’s optimization effect on improving vehicle utilization and reducing cost.Firstly,from the taxi booking service mode, analyze the behavior characteristics of the taxi and the user in taxi reserve,study feasible reserving taxipooling model. Analyze the cost structure of the mode system and put forward three kinds of relation models of unit mileage costs and capacity. Determine the value range of every parameter according to the actual situation of taxi operation. Establish the vehicle routing model with mixed time window to optimize the service path of the taxis in the system.Secondly, analyze all kinds of algorithms to solve the vehicle routing problem, determine the idea of using genetic algorithm to solve the model. Based on the characteristics of the problem, design the corresponding genetic operators such as crossover and mutation. And combine the tabu search algorithm and genetic algorithm, improving the global optimization characteristic of the algorithm.Finally make Harbin city taxi GPS data for example, generate user travel demand by recognizing the begin and end points as the input data of the model. Implement the algorithm with matlab, compare the optimization effect of the situation of taxi-pool and no taxi-pool and three different cost structure models on the result of optimization. At last, analyze the the parameter sensitivity. |
