A general inverse control model of a magneto-rheological damper based on neural network

Abstract

Because of the nonlinear hysteresis characteristics of the magneto-rheological damper, the damper’s inverse model has disadvantages of low fitting accuracy and poor practicality. Therefore, in this study, an optimized genetic algorithm has been proposed to optimize the back propagation neural network’s initial weights and threshold. Compared with other damper controllers, the proposed inverse model improves the control current’s prediction accuracy and tracks the desired damping force in real time. Moreover, the proposed inverse model and designed fuzzy controller are applied to the 1/4 vehicle suspension system simulation. The obtained results show that the optimized neural network model can be applied to a practical control. The root mean square value of body acceleration of semi-active suspension is lower than that of passive suspension under different road excitation. This method provides a foundation for the accurate modeling and semi-active control of the magneto-rheological damper.