Actuator Fault Degree Identification And Fault-tolerant Control Of High Speed Trains

With the rapid development of high-speed trains program in China, the safe and reliable operations of high-speed trains have become the major problems. Among them, the traction and braking actuator control and fault tolerance are the key sections in the traction and braking system and the safety of the train operations. Advanced control scheme is one of the safe and reliable technologies for enhancing the high-speed trains safety. On the basis of previous researches, the study is focused on traction and braking control of the high-speed trains and how to enhance the actuator reliability by using the active control schemes to construct the multiple safe layers to prevent possible fatal accidents. Then the paper establishes a dynamic model of the high-speed trains, which is more close to the reality, and proposes a train traction and braking actuator fault degree identification model. Based on fault degree identification signal, the reconfigurable fault tolerant control schemes are adopted in this paper to ensure the normal and reliable operations of the train speed and displacement tracking. The main work of this paper is as follows:In order to ensure that the faults of the high-speed trains are identified and in advance to use the actuator fault tolerance reconfigurable controller when the fault occurs, the paper is focused on how to achieve the actuator fault degree. Since actuator failures are usually accompanied with significant change of the speed or/and other related system states, the actuator fault identification can be considered as a regression problem based on support vector machine (SVM) where different states can be distinguished before and after actuator fault occurs. Regression model is trained by the sample set of input and output data of different fault degree. Then input a data sample set that needs to be identified, we can obtain the actuator fault degree by the result of regression, so as to realize the fault tolerant control of the reconfiguration controller. The practical function of fault degree identification method is proved in MATLAB experiments.Based on the complex environment, time-domain dynamic model of high-speed trains is established in the work. The model is multiple point-mass with single-coordinate dynamic model. And it includes the transient impacts, environmental resistance, interaction forces between the adjacent vehicles and actuator failures. With the multi-mass features of high-speed trains, the model has simple structure and the low dimension. Because the model is established, the effective control strategies could been applied to high-speed trains.At the same time, the problems of high-speed trains speed and displacement tracking after the failures of the traction and braking actuator are studied. When the actuator fault degree recognition model is used to identify the fault degree of actuator, the degree is used as the reconstruction factor of the reconfigurable fault-tolerant controller. Thus, reconfigurable fault tolerant control is realized. The schemes based on neural network can approximate the nonlinear parts of the train, automatically generate intermediate control parameters, neural network weights and compensation signals, and obtain traction and braking force input. This is the process of fault tolerant control based on neural network. Then the reconfiguration factor is added to realize the reconfigurable fault tolerant control based on neural network. In addition, the robust adaptive control is compared with the reconfigurable robust adaptive fault tolerant control. By using Lyapnov function to analyze and verify the stability of the system, the simulation results of MATLAB also prove that the fault-tolerant control methods can realize the control targets of actuator failures. Reconfigurable fault tolerant control has smaller displacement, velocity tracking error and better tracking performance.