A Neural Network-based Precise Parking Control Method For Urban Rail Trains

As an important part of public transportation system,urban rail transit is the main part of 3D urban construction and an effective measure to solve urban traffic congestion.Precise control of the running time of urban rail transit ensures the safety,punctuality and efficiency of train operation,which is conducive to better implementation of the operation scheduling plan,thus improving the service level of urban rail transit,and improving the operation efficiency of urban rail transit network.The running process of the train is uncertain,and the traditional PID(Proportional?Integral and Differential)control has its own limitations.It cannot make timely adaptive adjustment to the changing conditions,and it's difficult to deal with the complex operation environment by a single control algorithm.In view of the disadvantages of traditional train stop control methods,this paper starting from the precise parking requirements of urban rail trains,combines mathematical statistics,theory and model simulation,proposes a precise train stop control method based on neural network.First of all,according to different running lines and train s' conditions,the paper analyzes the train's traction,braking force and resistance in the running process detailedly,reveals the train's running rules,builds the train's dynamics model,and provides theoretical basis and data support for the paper.Secondly,the paper makes optimization and improvement on the basis of existing PID control,and proposes a new control method by combining neural network with PID control.From the perspective of system identification,the paper transforms the train brake model into abstract mathematical model,uses MATLAB/Simulink for model simulation,the simulation results show that the new control method has the ability of self-learning and self-adaptation,enhanced the anti-interference of the control system,improved the real-time control precision of the train operation compared with the existing PID control algorithm.Aimed at the uncertainty in train operation control system,this paper proposes the adaptive predictive control method of prediction error based on the grey prediction theory,combined with neural network PID control method,the grey prediction neural network PID control model is established,the simulation results show that grey prediction neural network PID control method can achieve better parking control accuracy,has a good support and reference for the determination of the train operation strategy.