Controller Design Of Magneto Rheological Semi-active Suspension Based On Inverse Model

With low cost, low energy consumption, simple structure, and goodperformance on ride comfort control, semi-active suspension has wideapplication prospect in the future on ordinary cars. Magneto rheological fluiddamper is one of the important components of semi active suspension, due to itslow energy consumption, fast response speed and wide control bandwidth, all ofthese make it research focus on architecture, machinery and automobile fields. Inthis paper, we give a semi active suspension control method based the inversemodel to duel with the complex nonlinear characteristics of the MR damper.Firstly, we analyze the nonlinear dynamic characteristics of MR damperbased on the experiment data and related literature and give the comparison ofthe existing mathematical models of MR damper. The analysis results show that asingle polynomial curve can fit the damper characteristics of one direction andthe hyperbolic tangent model can reflect the dynamic characteristics of the modelwith static fitting parameters. All of the two models are with simple structure.According to the analysis results of the MR damper model, the1/4semiactive suspension model is established with damper force as the control variableand the structure of the controller which is based on the inverse model isascertained. On this basis, we give the design of neural network inverse model(NNIM) and polynomial inverse model (PIM) and the comparative analysis ofthem with different current frequency and vibration frequency. The results showthat the accuracy of NNIM is not so much high because of the number of samples.So, in this paper we select PIM to transform the desired damping force intothe control current.Through the analysis of two kinds of typical semi active suspension control method "Skyhook" and ADD (Acceleration-Driven-Damper), this paperimprove the ADD algorithm with the established PIM and the new algorithm usedesired damping force as the control variable. In order to test the performanceof the new algorithm in frequency domain, we design the1/4simulationenvironment of magneto rheological semi-active suspension in SIMULINK. Theresult shows that the modified ADD algorithm can overcome the defect of theoriginal algorithm of poor performance in low frequency domain. In order tofurther verify the damping performance of the modified ADD algorithm, wedesign the optimal controller and the simulation shows that the modified ADDalgorithm can achieve the optimal damping effect.Considering the coupling between the4suspensions and the influence ofpavement, we further analyze the damping effect of the established MR semi-active suspension controller in the whole vehicle. Firstly, we modified veDYNA(dynamic simulation software with high precision) to establish avehicle simulation system with MR semi-active suspension system. With thissystem we compared the performance of ADD and control method of this paperin three kinds of typical road conditions. The result shows that, the newmethod is superior to the ADD algorithm in reducing the body vibration.