| Programming and optimization of passenger train scheme is not ony the key workof the railway passenger operations, but also complex multi-objective programmingproblem. As China’s railway passenger dedicated line increasingly forming intonetwork, railway passenger transport enterprises will be gradually in accordance withthe market-oriented operation and passenger train scheme needs to meet the objectivesand criteria which showns a complex and dynamic trend. On the other hand,passengers’ increasing travel satisfaction put a higher demand on the passenger trainscheme programming and optimization model. In order to solve the multi-objectiveprogramming problem of passenger train scheme, an in-depth study is started from themulti-objective optimization theory and a novel effective and robust multi-objectiveoptimization algorithm based on quantum genetic algorithm is proposed. Furthermore,a multi-objective optimization model based on social, economic and market efficiencyand a passenger flow forecast model based on the ticketing data of China RailwayTicketing and Reservation System(TRS) are systematically proposed and combinedwith actual data to conduct applied research and model algorithm validation. Thesolution proposed in this paper has important academic significance and referencevalue in the development and optimization of in-depth study on passenger trainscheme.The innovation of this paper is mainly reflected in the following four aspects:1. An improved multi-objective quantum genetic algorithm (IMOQGA) isproposed by combining the basic quantum genetic algorithm with constrainedmulti-objective optimization theory, and its improvement strategy including theintroduction of several concepts such as the the rasterization archive colony,constraint violation, the probability of migration colony and quantum crossover andother operations. Furthermore, the algorithm is validated on convergence of thealgorithm, the distribution of solutions and performance of solving constrainedmulti-objective optimization by solving the constrained multi-objective optimizationfunction problems. 2. Based on the researching background of Passenger Trains Scheme andcomprehensive analysis of influencing factors on Passenger Trains Scheme, amulti-objective optimization model of Passenger Trains Scheme is built onmaximizing train economic, social and market efficiency, which is validated onsystematical integrity and operability. Particularly, A creative passenger satisfactionindex to establish passenger allocation model makes possible unifying the passengerflow assignment problem into social efficiency goal in multi-objective optimizationmodel.3. Bi-level characteristics analysis on the characteristics of railway by extractingpassenger traffic data, a new Bi-level orthogonal neural network (BLON) modelbased on BP neural network is proposed. Substantially, the idea of the BLON model isfirst to use the relatively independent model to forecast on input characteristics, thenthe virtual structure named projection layers are added between the hidden layers andthe output layers by Gram-Schmidt transform to reduce redundant networkconnections in the model training process, finally the independent outputs wereincorporated to obtain the prediction result. This model is applied to forecasting theOD traffic volume.4. Research achievements about mentioned models and algorithm are actuallyapplied to Passenger Train Schema development and optimization. According to ODpassenger and the concept of node important degree to design initial passenger trainscheme, the improved multi-objective quantum genetic algorithm are applied topassenger train scheme optimization and achieve to directly solve the passenger trainscheme optimization model. As a practically application, a passenger train scheme ofthe railway corridor network around Beijing-Shanghai passenger dedicated line in2015is programmed to verify that the proposed algorithm can reasonably solving andoptimizing the passenger train scheme problem. |
PDF Full Text Request