A unified MR damper model and its inverse characteristics investigation based on the neuro-fuzzy technique

Abstract

Magnetorheological (MR) damper is one of the key technologies in the field of semi-active suspensions. However, the dynamic model and control strategy are not yet well established. A unified MR damper model and its inverse characterization are proposed to investigate MR damper dynamics. First, a novel unified model is proposed to describe the shear term of the MR damper force. This forward unified MR damper model uses one set of parameters to describe all MR force behavior under all currents and velocities, which agrees well with experimental data. Then, an inverse MR model based on the adaptive neuro-fuzzy inference system (ANFIS) technique has been established, where the precise training data are obtained from the proposed unified hysteretic model. A clear step-wise modeling process is proposed to decrease coupling degree of system which can be beneficial to improve inverse modeling efficiency. The inverse MR model is used for a semi-active suspension control, which can obtain the excitation current, exactly tracking the desired damping force computed by suspension control algorithm. Finally, validation of the proposed ANFIS inverse model has been conducted by comparison with three traditional empirical inverse approaches. The results reveal that the ANFIS inverse model could achieve much more tracking capability for the desired damping force and thus contributes to the development of semi-active suspension controllers.