Property Adjusting of Magneto-Rheological Damper Using Neural Network Adaptive Filter

Abstract

Magneto-rheological (MR) damper is one kind of damping changeable component which output damping force can be controlled by input current under certain working condition. There are many models have been developed to predict MR damper's output include LMS force method which has been developed in our previous research. But most of model with unchangeable parameter which have been selected under certain condition in lab cannot adjust itself to adaptive actual working condition even under a small disturbance, such as temperature. To develop a method which can adapt different working condition, an adaptive filter has been studied based on neural network with BP update program. Furthermore to adjust MR damper's property to expected damping function, an adaptive inverse control method has been presented with the adaptive filter in this paper. The numerical simulation results indicate that the adaptive filter can fix the error introduced by varied working condition, the control precision of characteristic of magneto-rheological damper is better than phenomenon model, and the damping characteristic of MR damper can be controlled by using the adaptive inverse control method.