Study On The Optimization Of Rolling Stock Planning And Crew Planning

Because of the relation between demand and supplyment and the structure of resource are changing continueously, the rolling stock planning problem and crew planning problem of passenger dedicated lines(PDLs) in our country display different characteristices from existed research and the characteristices change dynamically, which request the methodes to be more flexible and adaptable. Considering the similarity of these two problems, this paper classifies these problems with optimization problems of conveyance and crew planning problems in public transport as one type problem named optimization problem of movable transportation resource. By analysing and describing the commonness of this type problem, this paper proposes a general optimization method of this type problem. On the basis, this paper study on the optimization method of rolling stock planning problem and crew planning problem according to the operation characteristics of PDLs in our country,. The main work of this paper can be concluded as follows:1. By analysing the characteristics and inherent relations of optimization problem of movable resource in each field, this paper concluds the commom characteristics of these problems are building mapping relation between movable resource and transportation task and builds a relation function to describe these common characteristics. From the aspect of relation of time and space of problems, this paper also discusses the descomposition of problem.2. Aiming at the common characteristics and difference of this type problem, this paper builds time and space unifying network to describe the complex relation between time and space of problems, and hides different relation between time and space of concrete problem into specifical network structure. Using the time and space unifying network, this paper transforms the problems into time or space problems to decrease the dimension of problems. On this basis, this paper builds optimization model based feasible solution, designs algorithms framework based column generation, improves the generalized lable algorithm which is used to solve pricing problem, and concludes the using of the general oprtimization method as determination of five parameters.3. Based on the analysis and discussion of related parameters of rolling stock planning problem in the background of PDLs in our country, this paper bounds this problem. On the basis of general optimization method, aiming at concrete feature of problems, this paper studies the optimization method of rolling stock circulation and maintenance routing of train units in the "line planned by rolling stock" stage and in the "line adjusted by rolling stock" stage, respectively.4. Aiming at the characteristic of rolling stock circulation in "line adjusted by rolling stock" stage that its relation of time and space is complex, according to the speciality of time window of timetable, this paper describes the influence of time window to neighor trains by building connection network, describes the relation of time window among related trains by proposing the concept of mutually exclusive arc and proposes the method of judging feasible routing considering the relation of time window among trains in a routing of rolling stock. On this basis, this paper builds optimization model of problem and designs algorithm based artifitial variables.5. Aiming at the characteristic of crew planning problem of PDLs in our country, which has comlex and various crew rules and strong coupling of plans, this paper projects crew scheduling problem on time dimension using the abstract of connection network. Using the improved generalized labeling algorithm's advantage in deal with complex constraints, considering the relation of crew scheduling and crew rostering, this paper builds optimization model and designs algorithm considering the characteristic of problem. According to the speciality of crew rostering problem, this paper proposed the concept of equal combination of crew pairing and based this concept proposes heuristic algorithm.