The Optimization Theory And Method For Configuration Of Transportation Facilities In Urban Rail Transit Station With Consideration Of Stability Analysis

Due to the large-scale development and long-term overload operation of China’s urban rail transit system, it is now necessary to have the analysis methods for the changes of passengers accumulation in urban rail transit station and also system optimization theory for passenger facilities of urban rail transit station is required to account for the changes in passengers accumulation.Its essence lies in the stability analysis of the system with a large number of passengers and various types of transport facilities, and the establishment of the system optimization theory considering the stability for passenger facilities of urban rail transit station. However, the stability is not considered in the current research on optimization of passenger facilities, and there is a lack of appropriate methods to describe the stability.These problems can be solved by the use of queuing network modeling, QBD process theory, Phase-type (PH) fitting, queuing network sensitivity analysis, queuing network simulation, genetic algorithm and scatter search algorithm. Therefore, this thesis is related to the research on PH queuing network modeling, description of performances including stability, the optimization model and algorithm design for the station system with a large number of passengers and various types of transport facilities. In detail, the following is the main research work:(1) The description of the station queuing network system. The thesis builds a station queuing network system structure with batch queuing system A/B[0,C1]/1/C parallel queuing system A/B/C1/C, hybrid queuing system A/B/C1/C and state-dependent hybrid queuing system A/B(n)/C/C. This queuing network system accounts for the lack of considerations on capacity limit and state-dependence in the current research. Similarly, queuing phenomena for non-straight corridors, escalators and stairs are also described.(2) PH fitting and its parameters calibration for arrival interval and service time of the station queuing network system. Based on a large number of actual data, arrival interval and service time of the station queuing network system are fitted precisely by phase-type (PH) distribution. According to Code for Metro Design, the corresponding calibration method of PH distribution parameters is given for facilitating the practical application. Meanwhile, other common distributions are also used to fit the actual data in comparison with PH distribution. The results show that PH distribution best fits the arrival interval and service time in most cases, while the fitting of other distributions for arrival interval and service time that were used in past studies are not suitable.(3) The establishment of a PH queuing network model for the transport facilities of urban rail transit station. Based on the good analyticity of PH distribution and Quasi-Birth and Death (QBD) process, PH/PH[0,C1]/1/C bulk service queuing model is established for waiting trains, PH/PH/C1/C parallel queuing model is established for ticketing and ticket checking, PH/PH/C1/C hybrid queuing model is established for taking escalator, and PH/PH(n)/C1/C state-dependent hybrid queuing model is established for passing through corridors and stairs. Improved Successive Over Relaxation (SOR) algorithm with linear complexity is designed to solve the above PH queuing models. MATLAB numerical analysis, Java Modeling Tools numerical simulation and Any Logic microscopic simulation are applied for validating the results of models. Then, based on the above models, PH queuing network model is established for the transport facilities of urban rail transit station. Fixed-point iterative algorithm with linear complexity is designed to solve the above PH queuing network model. MATLAB numerical analysis, Java Modeling Tools numerical simulation and AnyLogic microscopic simulation are applied for validating the results of network model.(4) Quantifying of the performance indicators of station PH queuing network system. Based on queuing indicators of the above model, existing performance indicators of station are nondimensionalized and normalized by the method of traffic utility function for reflecting the aggregation state of passengers within the station. Besides, the stability indicator are nondimensionalized and normalized using elastic coefficient of existing performance indicators on queuing network system parameters and by the method of traffic utility function for reflecting the aggregation state changes of passengers within the station.(5) Establishment of the optimization model and the algorithm design for the configuration of station transport facilities. The stability indicator is included in the objective functions of the optimization model, and all performance indicators are deduced by PH queuing network model. Based on Convexity-Concavity of performance indicators and Discreteness-Continuity of optimization parameters, a global optimization algorithm, SS&IAGA, is developed to solve the optimization model. According to the actual situation, various types of numerical examples are presented to validate the rationality and necessity of the configuration method.In this thesis, the above studies provide basic theoretical support for the configuration, operation and management of transport facilities in urban rail transit station.