Study On Compiling Method Of High-speed Railway Line Planning Based On Machine Learning

The line planning of high-speed railway,compilation methods and optimization of which have always been the research hotspot,is the foundation of its operation as well as the basic carrier for matching transportation capacity to demand.In the process compiling and optimization of line planning,the existing research focus classical mathematical op-timization method and modern evolutionary method on modeling and solving it.However,these problems have complex constraints and many objectives,which make the algorithm more complex and computational complexity.At present,machine learning research has been widely used in various fields,which is suitable for solving data mining problems.In addition,the continuous accumulation of high-speed railway data provides a new way to solve the problem of line planning by machine learning.Therefore,this paper proposes a set of line planning compiling method for high-speed railway based on machine learning.The main research work includes the following four aspects:(1)The general solution framework of high-speed railway line planning compilation is constructed.This paper further describes the problems to be solved by combining the practical and theoretical processes when compiling the line planning.According to the description of the problem,the hypothetical conditions required for the study are put for-ward,and with machine learning as the objective function,the line planning model of high-speed railway under wide constraints is established.Furthermore the solution frame-work of the line planning of high-speed railway is put forward.(2)The evaluation model of high-speed railway line planning based on machine learning is established.Seat occupancy rate is selected as the key index for evaluating the line planning by analyzing the related indexes of high-speed railway.The input data fea-tures suitable for machine learning are constructed through three steps:feature extraction,data standardization and feature selection.Secondly,according to the characteristics of high-speed railway line planning with high dimension and time sequence,LSTM deep learning is selected.The model is used as the evaluation model,and the evaluation model of high-speed railway line planning based on LSTM is designed.Finally,the algorithm of training LSTM model using high-speed railway line planning and seat rate data is stud-ied,and the performance evaluation of the model is further optimized.(3)This paper presents a method based machine learning for high-speed railway line planning.Because the initial feasible set of high-speed railway line planning has great influence on the search process,an initialization method for high-speed railway line plan-ning is designed,which reduces the search space of the algorithm and improves the start-ing point of the algorithm.By clustering method,the set of reserved trains is calculated,so that it remains unchanged in the iteration calculation,and the crossover and mutation genetic operators for variable train parts are designed.Then,the algorithm flow of high-speed railway line planning compilation based on machine learning evaluation model and genetic algorithm is given,and the feasibility and convergence of the algorithm are further analyzed.(4)A case verification experiment was carried out for the Beijing-Shanghai high-speed railway line planning.Firstly,the existing background of the Beijing-Shanghai high-speed railway is described,and the data and experimental environment are prepared.The methods used in the experiment are specified.The method of high-speed railway line planning based on machine learning is verified by experiments,and the results were ana-lyzed and discussed.