A recurrent neural network emulator for the inverse dynamics of an MR damper

The magnetorheological (MR) fluid damper is a relatively new type of device that shows future promise for control of vibration. The dynamic behaviors of this nonlinear semi-active damper has been modeled using a Bouc-Wen hysteresis model. This model estimates damper's forces based on the inputs of displacement, velocity and voltage. However in some control applications, it is necessary to command the actuator so that it produces desired control forces calculated from some optimal control algorithms. In this study, we explore such a possibility via the neural network (NN) technique. A recurrent NN model will be constructed to emulate the inverse dynamics of the MR damper. To illustrate the use of this NN model, an optimal prediction control of an SDOF system is presented. Numerical results indicate that, via the help of the NN technique, we can command the Forces of the MR damper so that they follow closely with those desired optimal control forces. It appears that it is possible to utilize the MR damper in a more "optimal" and "active" way