Sectional Multi-objective Optimization Algorithm For The Urban Train Operation

With the rapid development of urban railway transit, the train contributes a more and more significant importance as it is a major service provider for passenger traffic transportation. Safety, punctuality, energy consumption, riding comfort and stopping at adequate position are the important assessment criteria for train operation. In this paper, a modified co-evolutionary algorithm is proposed for the optimization of train operation.Firstly, the mathematical model of the train operation, what the multi-particle train model and the complex track line conditions, such as ramps, curves, limited speed, what was also considered, was described. With the Multi-objective theory and method, the train operation optimization model aimed at energy saving, on-time, stopping at adequate position and riding comfort was established with safety constraint.In this paper, in order to improve the search capacity of the co-evolutionary algorithm as it is applied in the train operation optimization, the multi-objective co-evolutionary algorithm (MOCEA), based on the idea of competition and cooperation, has been improved. Through the introduction of the concept of dominance and improved the evolutionary genetic process of MOCEA, the co-existence co-evolutionary algorithm (CCEA) was proposed. The comparison between the two algorithms was made with the test function for multi-objective optimization algorithms, and the superiority of the CCEA was verified.As the train running in the urban railway transit has a relative constant operation strategy in part operation phases, the interval between two stations was divided into departing section, deceleration ramp, acceleration ramp, constant-speed ramp and arriving section based on the traction performance of a train and the cross section and speed limitation of the line. The initial operation strategies of the train in all the sections were initialized based on the operation experiences of the excellent drivers. A strategy of a section was seen as an individual, and the individuals in the same section constitute a population. The preference operator was designed according to different optimization objectives of different sections, and led the populations to evolve toward the expectation. The evolving process was divided into three stages, and the parameters of the self-learning operator changed with the stages in order to speed up the convergence of the populations. In order to increase the feasible solutions, the selection and variation of the individuals went on with the restriction of the safety, punctuality and the precising stopping. Then, the populations were further evolved with the co-existence operator and the self-learning operator of the CCEA algorithm, and the optimal operation strategies within all restrictions were finally generated.Lastly, Xi’an Metro North railroad passenger station to station operating range Wei Qu, for example, the optimization process of the train running simulation analysis, optimization obtained satisfactory results.