Study On Models And Algorithms Of Dynamic Fleet Management Problem

Organization and optimization of freight transportation operation is the important way and the key of decreasing transportation cost and increasing transportation efficiency. The vehicle is the carrier of the freight and the most important used resource in the process of freight transportation. The core of organizing and optimizing freight transportation operation is allocating reasonably vehicle to transportation task to eliminate some unreasonable instances so as to realize the object of using the least cost to get the most income. Under the background and demand, fleet management problem is brought forward. Managing effectively freight vehicle can elevate the efficiency of using vehicle and realizes scientific freight transportation. And to research generally the fleet management problem is the base and the key of constituting perfect freight transportation system, establishing modern scheduling system, realizing scientific logistics and improving intelligent transportation system.This paper deeply analyzes the shortcoming in the traditional literature. And the issue on the dynamic fleet management problem is fully discussed from some factors affecting vehicle allocation, vehicle allocation form, establishing vehicle allocation scheme and dynamic property of the problem. To know the run theory of the problem, the operation process of dynamic fleet management is fully analyzed. On the basis of the above work, dynamic fleet management problem is classed as deterministic problem with homogeneous vehicle type, deterministic problem with multiple vehicle types and stochastic problem. And the model and algorithm of the dynamic fleet management is fully researched.To the deterministic dynamic fleet management problem with homogeneous vehicle type, a particular linear function is devised to approximate the recourse functions from next time period to the end of the planning horizon through the use of the approximation techniques. The new-style model is formulated. And the dynamic fleet management problem by time and space is decomposed into a series of local problems by the linear approximation, one for each terminal. The local problem is a simple sort problem.To the deterministic dynamic fleet management problem with multiplevehicle types, the methodology to handle heterogeneous fleets of vehicles with flexible substitution of different vehicle types for different types of tasks is analyzed. Then the model of dynamic fleet management problem with homogeneous vehicle type is extended to handle heterogeneous fleets of vehicles. The specialized algorithms are devised to find a near optimal solution of the local problem of dynamic fleet management problem with multiple vehicle types.To the stochastic dynamic fleet management problem, the stochastic character is analyzed. With probability distribution function of random variable denoting the task numbers, some approaches are devised to approximate expected vehicle supply, and computer the vehicle dispatch probability and the expected contribution of the vehicle. Then the slope of the linear approximation function can be computed in virtue of above results. The linear approximation function uncouples the problems at different terminals so that we arrive at the local problem for each terminal at each time period.