Modeling of Magnetorheological Damper Using Neuro-Fuzzy System

Abstract

MR damper has so strong nonlinear characteristics owing to the unclearnonlinearity of the MR suspension in it, that it is very difficult toexpress the direct model. What's more, it is much more difficult for theinverse model. The paper puts forward a novel train of thoughts toidentify the inverse model of the MR damper considering the universalapproximation ability of fuzzy system. A neuro-fuzzy system is designedto identify the inverse model of MR damper based on Adaptive Neuro-FuzzyInference System (ANFIS). The ANFIS system is similar to its physicalcounterpart of the MR damper with the same number of the inputs andoutputs. The numerical simulation demonstrates that proposed neuro-fuzzysystem can accurately identify the inverse model of the MR damper forthe training data, and well approximate for the checking data. This ideacan be also used to model and control other MR damper with its directmodel unknown.