Research On Lateral Active Suspension Control Of High-speed Train Based On Repetitive Learning

In recent years,in the context of the rapid development of China’s high-speed railways,not only the speed of high-speed trains has been greatly improved,but railways have also been built in many severe locations,which has brought great challenges to suspension systems,especially for the suppression the lateral vibration.The limitations of passive suspension systems are becoming more and more obvious,and active suspension control is a powerful means to improve the lateral stability of trains.Therefore,it is of great practical significance to propose an efficient and practical active suspension control strategy to control train suspension systems.Firstly,the development of active suspension control strategies is briefly introduced,as well as the repetitive learning control(RLC)strategy used in this paper,including its control basis and research situation.Different from the existing control strategies that cannot use the periodicity of lateral dynamics to suppress the lateral vibration of trains,RLC can transform the lateral vibration suppression of trains into a tracking problem for unknown periodic disturbances or unknown system dynamics with periodic characteristics.This part is the control basis and theoretical basis for the controller design.Secondly,a 3 degrees-of-freedom(DOF)lateral active suspension model is used for controller design and a 17 DOF lateral active suspension model is used for simulation.The system input,i.e.,lateral irregularities,is divided into two types: deterministic periodic lateral irregularities and random lateral irregularities,and the corresponding numerical simulations are performed.Under different input conditions,the principle of determining the repetitive learning period is proposed.For the former,the repetitive learning period is deterministic and set to 1 s.For the latter,by spectrum analysis,an important conclusion,i.e.,the suspension system mainly counteracts lateral irregularities at 0.5 Hz,and mainly suppresses the lateral vibration at 0.5 Hz,is arrived,so the repetitive learning period is set to 2s.In the design of the repetitive learning controller,a periodic repetitive learning term that can continuously learn and improve the lateral vibration of the vehicle body is included,and the learning convergence of the entire closed-loop system is proved by Lyapunov stability theory.By introducing the vector synthetic lateral vibration index,the designed controller is converted and applied in a 17 DOF lateral active suspension model,and the RLC-based lateral active suspension system is designed.By introducing the evaluation indices of vibration reduction performance,the multi-objective optimized cross-entropy algorithm is utilized to optimize the selection of repetitive learning controller parameters.Finally,the RLC-based lateral active suspension system is compared with the lateral active suspension system based on dynamic matrix control(DMC)system and passive suspension system.Under both types of lateral irregularities,the simulation results show that the RLC-based lateral active suspension system has better suppression performance for lateral vibration,especially for low-frequency vibration in the frequency range of(7),]0 3 Hz,which is also the human body’s most sensitive frequency range for lateral vibration.Further simulations under different speed conditions show that the repetitive learning controller does not need to adjust the control parameters and learning cycle online in real time,but also maintains excellent suppression performance for lateral vibration and has certain robustness.