Neural Network Adaptive Fault-Tolerant PI Control For High Speed Train

With the development of social economy,the demand for traffic transportation is also increasing.High speed railway becomes an important part of the transportation system gradually,which is not only an inevitable requirement for the development of the national economy,but also an important support for the implementation of the national strategy.Therefore,the stable and safe operation of the high speed train is related to the safety of the passengers' life and property.However,the study of train operation safety theory is still at the initial stage,and there are not enough researches on the possible actuator failures of the traction/braking system in the long running process,which has not formed a complete theory system.In order to ensure the safe and efficient running of the high speed train,it is necessary to study the safe operation of the high speed train.The control of traction/braking actuator system is the key point to the safe operation of high speed train.When the speed is greater than 300 km/h,once the fault occurs,the consequences would be serious.Targeting at the possible actuator failures of high speed train,this dissertation studies the adaptive PI(Proportion Integral)control strategy based on neural network to maintain the safe operation of high speed train under the condition of sudden failures of actuators.The main work and contribution are summarized as follows;(1)In this dissertation,the aerodynamic resistance law under high speed condition is analyzed.Based on the theory of dynamics,the physical model and dynamics equation of single mass point and multiple mass point of high speed train system are established.At the same time,considering the model uncertainty,system nonlinearity and external disturbance,a time domain model of "multiple mass point single displacement" is used.(2)Considering partial or total failure of actuator,a self-adaptive PI control strategy with simple structure and low computation cost is designed for high-order nonlinear MIMO system.In addition,the selection of parameters does not require the repeat trial process as the same as traditional PI controller,and the controller parameters can be updated adaptively according to the algorithm.Based on Lyapunov stability theory,the stability analysis of adaptive PI fault tolerant control strategy is carried out.Finally,the effectiveness of the control strategy is verified by MATLAB numerical simulation,which shows the method has strong fault-tolerant ability for actuator faults,and is robust to unknown time-varying characteristics,nonlinearity and external disturbances of the system.(3)Based on the idea of iterative theory,a simple iterative algorithm is designed.The accuracy of tracking control for MIMO system is further improved,and the effectiveness of the iterative algorithm is verified by MATLAB numerical simulation.(4)Considering the ability of RBF(Radial Basis Function)that it could approximate continuous nonlinear function by any given precision,construct the adaptive PI controller based on RBF neural networks.The control strategy does not depend on the accurate system model,with convenient operation and simple structure.MATLAB is used to simulate the possible actuator faults during the train running process.The result shows that the neural network adaptive fault-tolerant PI control strategy can realize the accurate tracking control of the high speed train position and speed with good robustness.