Research On Energy-efficient Timetable Optimization For Urban Rail Transit Line Based On Data-driven Methods

Urban rail,which plays an irreplaceable role in public transportation due to its punctuality,safety,comfortable,low energy consumption and high load capacity.Nowadays,with the largescale construction of the rail network and the increasing number of passengers,energy consumption and operational cost are increasing.As the traction energy consumption is an important part of the total operational consumption,how to effectively reduce the traction energy consumption is the key to quickly solve the total operational consumption and is also an effective measure to save system costs.A precise prediction of traction energy consumption is beneficial to the establishment of operation organization mode and the evaluation of energy efficiency,thus decreasing the traction energy and system costs.Moreover,optimizing the train energy-saving timetable is also an efficient means to reduce the operation energy and cost.As a significant guide for operating schedule,the timetable is closely related to the train operation mode.Therefore,optimizing train speed and energy consumption can optimize an energy-efficient timetable.The main purpose of this thesis is aiming to optimize energy-saving timetable based on data-driven with analysing the factors,principles,and measures of energy consumption.Then,real data from the practical data is applied to propose effective energy-saving strategies from two aspects of energy evaluation and dispatch operation by using machine learning.In detail,the main contributions of this paper are as follows:(1)In this study,we provide two prediction models of traction energy consumption based on machine learning.Because the factors affecting the traction energy consumption of trains are complex,and the traditional mathematical simulation and regression method are difficult to ensure the prediction effect.In view of this,a method for predicting traction energy consumption based on machine learning is proposed.Two machine learning methods,Support Vector Regression(SVR)and Random Forest Regression(RFR)are utilized to establish the forecasting model of train traction energy consumption.Firstly,six typical factors are selected.And,influences from both single and multiple factors are analysed.Then,the optimal parameter combinations are searched with an enumerative method.Finally,the proposed method is verified by taking the real operation energy-consumption data of the Beijing Metro Changping Line as an example.The results show that both SVR and RFR are stable and can achieve high prediction precision.In addition,the RFR model is utilized to rank the importance of factors influencing traction energy consumption,which provides a strong focus and reference for energy-saving strategy evaluation of transportation systems.(2)In this study,we provide two energy-efficient timetable models to minimize energy consumption and total operational cost.Since passenger quality and train speed have an influence on traction energy consumption,a Mixed-Integer non-linear Programming(MINLP)model is proposed,including the non-linear objective and constraints.To find the exact solutions,the objective and constraints are linearized.Then,a Mixed-Integer Linear Programming(MILP)model is employed,which can be solved using the commercial solver Gurobi.Furthermore,from a view of system costs,we present an alternative objective to optimize the total operational cost.Real Automatic Fare Collection(AFC)data from the Changping line of Beijing urban rail transit is applied to validate the model in the case study.The results show that the designed timetable could achieve about a 35% energy reduction compared with maximum energy consumption and a 6.6% cost-saving compared with the maximum system cost.