Model Predictive Control Of Active Suspension Based On Road Information Recognition

Suspension system is an important part of vehicle,which determines the ride comfort and handling stability,and plays a great role in improving the overall performance of the vehicle.Suspension system can be categorized as passive suspension,semi-active suspension and active suspension.Among them,active suspension can flexibly deal with different road conditions to obtain the best performance,so it has become a hot spot in the automotive industry research and development.In this paper,the active control of vehicle suspension is studied based on model predictive control and road roughness identification.The paper mainly includes:(1)Establishing the suspension model of quarter vehicle,half vehicle and whole vehicle,the mathematical model of permanent magnet synchronous linear motor actuator and the road excitation model.The performance evaluation method of suspension system is elaborated.The results provide a foundation for road identification based on vehicle response and control of active suspension.(2)Using NARX neural network to identify road roughness in time domain.Bat algorithm is added to optimize the parameters of NARX neural network to achieve the best recognition effect.The influence of different road grades,different speeds,random noise and different sprung masses on the recognition results is analyzed,which proves the effectiveness and applicability of the neural network method to identify road roughness.(3)Combined with recognition of road roughness,the model predictive control of active suspension is realized based on one quarter suspension,half vehicle and seven degree of freedom vehicle.The influence of parameters and preview mode on the predictive control effect is studied.For the disadvantage of model predictive control which depends on modeling accuracy,the feedback correction method is used to improve its robustness.For the disadvantage of large amount of online calculation of model predictive control,explicit model predictive control is used to improve its timeliness.The validity and correctness of the method are verified by simulation.(4)Designing and building a quarter vehicle active suspension bench test platform.The road identification and model predictive control of active suspension are verified by bench test.The experimental results show that road identification based on the neural network and active suspension using model predictive control are effective.