Multi-objective Optimization Method For Urban Rail Train Operation Based On Improved Differential Evolution

Due to the characteristics of great safety,huge passenger capacity,fast and convenient,high comfort,energy saving and great efficiency of urban rail transit,it has become an significant development target of my country’s urban public transport system.With the booming progress of urban rail transit,the sustained increase in operating mileage and passenger volume has led to a steady growth in operating energy consumption.Therefore,how to realize energy saving has turned into a research hotspot for green and sustainable development of urban rail transit.While,train operation control is a complex problem that integrates multiple goals of safety,punctuality,energy saving,comfort and accurate stopping.Thus,it is of extraordinary significance for the development of today’s urban rail train operation control to how to seek a "satisfactory" operation strategy from a large number of operation control methods that can meet the requirements of multiple goals.Aiming at the multi-objective optimization problem of urban rail trains,the crucial research work of this thesis is generalized as follows:Firstly,a longitudinal dynamics model of the subway train based on actual line conditions is established by analyzing the force situation of urban rail trains in the track direction under different operating conditions.Combined with the train dynamics model,a multi-objective operation optimization model of subway train is established to achieve the train safe,punctual,energy-saving,accurate parking and comfortable operation.Compared with the traditional Gaussian comfort model,the novel comfort index combined with the bell-shaped model is constructed,which can more accurately describe the comfort level and thus better improve.Secondly,to address the disadvantages that traditional DE algorithm fall into local optimum easily and lead to slow late convergence when applied to practical engineering,two improved algorithms are proposed based on the solving method of multi-objective optimization problem.The improve the adaptive differential evolution(ADE)algorithm is designed by introducing simulated annealing idea,the initialization improvement strategy of inverse trigonometric function logistic mapping and the control parameter adaptive strategy.An improved multi-objective differential evolution(MOADE-MMS)algorithm is proposed by introducing non-dominated sorting and crowding calculation strategies,the integrated adaptive multi-mutation strategy and parameter adaptive update strategy.The optimization performance of the two algorithms is verified by the comparison experiment and optimization simulations from low-dimensional to high-dimensional adopting different algorithms for a variety of typical test functions is performed by using MATLAB.The results of simulation comparison prove that the improved algorithm has certain advantages in optimization accuracy and stability,which tested the validity and superiority of the improved algorithm and the feasibility and reliability of using the improved algorithm to solve the multi-objective optimization model in this thesis.Then,two approaches are proposed for the multi-objective optimization model of urban rail train operation.One is the traditional solution method converted to the single-objective optimization problem through simplified processing.The penalty function method is selected to comprehensively consider the multiple performance indicators to design the fitness function,and the improved ADE algorithm is applied to solve the multi-station operation strategy of multiple performance indicators optimized evenly.Another is to directly solve the complex multi-objective optimization problem based on the Pareto rule,and the improved MOADE-MMS algorithm is adopted to search the multi-objective train operation strategy.Finally,through using MATLAB as the experimental tool,a simulation experiment based on the practical application scenario of a Beijing subway line was carried out,and the results are compared and analyzed from the two dimensions of the algorithm and the approach,which verified that the algorithm and method proposed in this thesis can effectively reduce operating energy consumption and improve comfort.