| Summary: | 碩士 === 國立暨南國際大學 === 資訊管理學系 === 92 === The relocation problem, originated from a housing project in Boston area, is a generalized resource-constrained scheduling problem in which the capacity of a building after the redevelopment is not necessarily the same as that before the project started. In this thesis, we consider the scenario that several generalized due dates are specified to define the amounts of new housing units should be built in the whole project duration. Different from conventional due dates, generalized due dates are job independent and common to all jobs. In the problem of concern, each generalized due date is associated with an expected percentage of the completion of the project. At each generalized due date, if the actual percentage of completed housing units is greater than the expected one, a reward is given to the contractor of the project; otherwise, a penalty is incurred. Given an initial amount of temporary housing units, the goal is to find a feasible reconstruction sequence that maximizes the total reward around all generalized due dates. In our study, we present the computational time complexities of the general problem and a special case. Upper bounds and dominance properties are also proposed for the design of branch-and-bound algorithms. We conduct computational experiments to study the effectiveness of the proposed properties. Statistics from the experiments show that the proposed properties can significantly reduce the computation time required for composing optimal schedules
|
|---|
