Neural network modeling and controllers for magnetorheological fluid dampers

One of the challenging aspects of utilizing magnetorheological (MR) dampers to achieve high level of performance is the development of accurate models and control algorithms that can take advantages of the unique characteristics of these devices because of their inherent nonlinearity. In this paper, the authors proposed a direct identification and an inverse dynamic modeling method for MR dampers using recurrent neural networks. Based on the above neural network models, a configuration for the MR damper controller is also explored. The command voltage for the MR damper can be obtained through the neural network model according to the desired damping force determined from the system controller. The architectures and the learning methods of the direct and inverse dynamic neural network models for the MR damper are presented, and some simulation results about the MR damper controller are discussed.