A Study On The Optimization Method Of Maintenance Operation Plan In Depots Of Electric Multiple Units

In recent years,the development and construction of high-speed railways in China have made remarkable achievements.With the continuous completion of new high-speed rail line projects,the total number of Electric Multiple Units(EMU)trains that put into operation becomes increasingly large,which brings unprecedented pressure and challenges to the EMU depot maintenance staff.To ensure the maintenance work be completed on time with guaranteed quality in a well-ordered manner,the depot staff need to make depot maintenance plans in advance.The depot maintenance plan is a key document for guiding the maintenance work,and plays a vital role in coordinating train maintenance tasks,depot maintenance crew members and depot maintenance infrastructures.The depot maintenance plan design and adjustment process is challenging because it involves a lot of influential factors.Currently,the depot maintenance plan is mainly manually made by depot dispatchers.As a result,both the efficiency of plan development and the plan quality are significantly unsatisfactory.Furthermore,under the background that the fleet size of EMU trains becomes increasingly large and the maintenance workload become increasingly heavy,the manual method could not even provide a feasible plan within acceptable time.Therefore,it is of great significance to investigate the intelligent plan development methods as well as the plan optimization methods,for achieving higher efficiency of the plan development process,better quality of the maintenance plan,higher depot maintenance capacity utilization rate,and lower maintenance costs.At present,the existing literature on the depot maintenance plan optimization problem is not rich.Overseas studies focus primarily on the Train Unit Shunting Problem(TUSP)while domestic research pay more attention to the train maintenance process optimization problem.Moreover,since the train utilization strategies and maintenance policies in China differs from those in other countries,existing studies cannot be applied to this work directly.This thesis focuses on the depot maintenance plan optimization methods and studies primarily on three key ’sub-problems:the depot shunting scheduling problem,the depot shunting driver task assignment and route planning problem,and the depot maintenance crew scheduling problem,aiming to provide solid theoretical and decision supports for the depot dispatchers and maintenance staff.The detailed research work of this thesis can be summarized as follows:(1)The EMU train maintenance policies,the EMU train utilization and maintenance plan development processes and methods,and the EMU train maintenance base(especially the depot)maintenance planning process in current China high-speed railway system are analyzed.Based on the analysis results,three sub-problems of the depot maintenance plan optimization problems are extracted.For the three sub-problems,main impact factors,input and output are analyzed;and the objectives and main constraints that need to be considered are clearly explicated.Moreover,the relationship among these sub-problems are clarified;and the mathematical background of the problem is described,which lays a solid foundation for mathematical modeling hereafter.(2)The depot shunting scheduling problem is studied.After analyzing the typical workflow of EMU train first-level maintenance,a time-space network for the depot shunting scheduling problem is designed.Using this time-space network representation,binary decision variables as well as integer decision variables are adopted to capture the key elements of a depot shunting schedule.Then an integer programming model for the depot shunting scheduling problem is proposed.Consider that the original model contains a set of non-linear constraints,which could bring great challenges to solve it,a class of linearization techniques is proposed.After the linearization process,the original model can be refined into a standard integer linear programming model.Using the Gurobi solver,the computational results of a series of small-scale artificial instances indicate that the model can be solved to optimality within very short computational time,which validates the proposed solution approach.With the help of sensitivity analysis techniques,the model structure is anatomized;meanwhile,some interesting insights are revealed,which are beneficial to have a better understanding of the nature of the depot shunting scheduling problem.(3)The depot shunting driver task assignment and route planning problem is investigated.Firstly,by comparing with the Vehicle Routing Problem(VRP),a two-stage optimization framework is proposed for the depot shunting driver task assignment and route planning problem.The first-stage optimization model aims to obtain the optimal shunting driver size that is able to cover all the given shunting tasks.While the second-stage model aims to obtain the optimal shunting task assignment plan as well as the driver routing plan based on the shunting driver size determined in the first-stage model.In order to comprehensively consider the overall workload of all the shunting drivers and the workload balance level,the second-stage module is formulated as a bi-objective programming model.Then a modified normalized normal constraint method for obtaining the Pareto optimality solutions of bi-objective programming problems is proposed.A series of artificial instances validate the proposed method.Finally,based on the computational results of the artificial instances,a driver size estimation approach is suggested using the linear regression analysis,which provides a simple yet practicable decision support tool for the depot plan makers.(4)The depot maintenance crew scheduling problem is addressed.Using some research methods for the Resource-Constrained Project Scheduling Problem(RCPSP),an integer linear programming model is proposed for the depot maintenance crew scheduling problem.The objective of the model aims to minimize the maintenance project duration,and the constraints consider the uniqueness of the assignment of maintenance packets,the compatibility between the maintenance packets and maintenance crew types,the capacity of maintenance crew,the mutual exclusiveness of maintenance packets,the duration time of maintenance packets,and the maintenance time windows,etc.The proposed model is a general model that is able to apply to normal circumstances.To make the model be applicable to more complicated scenarios,various model extensions are discussed.Furthermore,the critical route of the maintenance project is analyzed using the Gantt chart representation.(5)Real-world case studies are carried out based on the Shanghai South Depot.Supported by the projects "Deep analysis of EMU train maintenance technologies-Study on the maintenance process optimization in the EMU maintenance bases" and“Deep analysis of EMU train maintenance technologies-Study on optimization for the EMU train utilization and maintenance plan" from the China Railway,using the real-world data obtained by field investigations,empirical case studies on the proposed solution approaches are conducted under the background of Shanghai South Depot.