Study on control strategy of magneto rheological semi-active suspension with neural network inverse model

In this paper, a neural network inverse model of magneto rheological (MR) damper is established and combined with a new force control algorithm to achieve the controlling of vehicle semi-active suspension. Combined with the Skyhook algorithm and ADD (Acceleration-Driven-Damper) algorithm, this paper presents a new damping force control algorithm which can improve the high frequency damping characteristics and is easy to combine with the current damper mechanical model. In order to realize the transforming from damping force to drive current, we further analyze the characteristics of the modified Bouc-Wen, polynomial and many different models, and combined with the test data, we get a hyperbolic model which can better reflect the MR damper dynamic. In order to facilitate the real-time calculation, a neural network inverse model is established based on the hyperbolic model. Finally, the improvement of the control method on suspension comfort performance is validated through the 1/4 suspension and full vehicle simulation.