Nonlinear Predictive Control For Automatic Train Operation With Consideration Of Safety Constraints And Multi-point Model

With the increase of train speed and operation density,higher requirements are put forward for the automation level of train operation.Therefore,Automatic Train Operation(ATO)has become one of the key research contents in transportation field.At present,significant progress has been made in the study of ATO control.However,the existing results describe dynamic characteristics of train inaccurately since they neglect dynamics of traction force and braking force.In addition,nonlinear characteristics become salient as train speed significantly rises,traditional linear models can no longer satisfy the design requirements of high performance controllers.Furthermore,control performance of the automatic operation system can be degraded without overall consideration of the cooperation of torque loop and speed-position loop.This dissertation focuses on ATO control problems.Firstly,dynamics of traction force and braking force as well as nonlinear characteristics of aerodynamics are analyzed.Moreover,according to uncertainties of train model and unknown time-variant characteristics of model parameters,the corresponding nonlinear multi-point dynamic models are established.Then,considering the unique safety constraints of train operation,and taking speed-position profile tracking performance,energy consumption and passenger comfort as control objectives,serval ATO nonlinear predictive controllers for speed loop and model predictive direct torque controllers for torque loop are designed correspondingly.Lastly,theoretical analysis of the feasibility and stability for the closed-loop system are given.The main contents and contributions of the dissertation are as follows:Firstly,the dynamics of traction force and braking force are considered and integer variables of train's running status and car types are introduced,thus a nonlinear multi-point dynamic model is built.Moreover,with the objective of electromagnetic torque and stator flux tracking,a single-step model predictive direct torque controller with one-step delay compensation for torque loop is designed.Then,a nonlinear predictive controller for speed loop is designed under consideration of safety constraints such as train collision,train derailment and off-hook cases.Theoretical analysis of the feasibility and stability for the closed-loop system are presented at last.Secondly,a nonlinear multi-point model of train with bounded disturbances is established after analyzing uncertainties of train model.Moreover,according to regional pole placement theory,a speed estimation algorithm is proposed for speed sensorless working condition.Then,a torque loop multi-step model predictive direct torque controller with one-step delay compensation is designed,and feasibility of the proposed algorithm is ensured by using the “deadlock”prediction method,convergence analysis of the torque loop system is given simultaneously.In order to avoid uncertain bounded disturbance that may cause the train to exceed the speed limit and fail to meet the safety constraint,a constraint tightening strategy is adopted to design a speed loop robust nonlinear predictive controller.Theoretical analysis of the algorithm feasibility and proof of stability for the closed-loop system are presented.Lastly,a nonlinear multi-point model of train with unknown parameters and bounded disturbances is established by analyzing the unknown time-variant characteristics of model parameters and the uncertainties of train model.According to speed sensorless working condition and time-varying motor parameters,a speed estimation algorithm with a certain degree of freedom is proposed based on regional pole placement theory.Moreover,an adaptive parameters estimation algorithm is proposed to improve accuracy of the predictive model.Then,a torque loop multi-step robust model predictive direct torque controller with one-step delay compensation is designed to deal with the uncertainty of traction motor model,feasibility of the algorithm and robust stability of the torque loop system are analyzed simultaneously.On this basis,a speed loop adaptive robust nonlinear predictive controller is designed by adopting tube MPC.Theoretical analysis of the algorithm feasibility and proof of input-state stability for the close-loop system are presented.