Ramp Control Model Of Metro Train And Its Energy-saving Control Based On Fuzzy Generalized Predictive Control

Building more accurate train model and better control of train operation under the condition of increasing train speed to reduce energy consumption has gradually become a research hotspot.According to the characteristics and running conditions of the actual running environment of subway trains,considering that the train running is not only restricted by its own running conditions,but also interfered by the external environment,this paper mainly aims at the modeling of the train,the following work was carried out in parameter identification and ATO system design:1.Firstly,considering the nonlinear,time-varying and even uncertain characteristics of the subway train model,the train is regarded as a single particle,considering the external environment disturbance,the running state of subway is divided into four parts,which are running empty,rising,limiting and compensating.2.Then,in order to obtain the parameters of the model more accurately and quickly,a new method is proposed to identify the parameters of the metro train in the low-speed stage,where the train can be approximated to a linear least squares.A parameter identification method of nonlinear system based on CNN-IGRU neural network is proposed in the middle and high speed stage,using CNN to extract the features of train data,replacing the activation function of GRU and optimizing the network structure,using the improved GRU to identify the model parameters,different constants are used as the first layer neuron weights to represent the train data,and model parameters are used as the second layer neuron weights to train and complete the identification.3.Finally,aiming at the energy-saving ATO of metro train,in order to overcome the system model error and external disturbance,a fuzzy generalized predictive controller is established,the deviation between the desired target and the actual output and the rate of variation of the deviation are taken as the input of the fuzzy controller,and the output of the fuzzy controller is taken as the compensation term of the generalized predictive controller.At the same time,considering the different control objectives in the traction/cruise and braking stages,the train adopts the traction model to track the optimal speed curve in the traction/cruise stage in order to realize the energy-saving and efficient operation,track the distance curve during the braking phase to achieve accurate parking.The validity of the parameter identification method,the veracity of the model and the superiority of the controller are simulated by taking the Beijing Yizhuang subway train as an example,the simulation results show that the parameter identification model of CNN-IGRU neural network is faster and more accurate than LSTM and GRU model.The controller can track the given velocity-distance curve accurately,and has the fast response ability.It is a reference to the design of ATO energy-saving controller.