Kouhi S., Montazeri A., Poshtan J.

The use of adaptive recurrent neural network for active vibration control of a laminated thin plate is presented. The various robust control approaches applied for active vibration control requires a MIMO model of the plate and exact knowledge of accompanied uncertainties to attain a good trade-off between robust stability and robust performance of the controllers. In contrast, the powerful learning capabilities of recurrent neural networks to capture the dynamics and nonlinearities of the system for identification and control purposes will be a good choice. For this purpose two neural networks, one for identification and one for control purpose is utilized and they are adapted to conform with the changes in the environment and possible uncertainties that may occurs. The results are also compared with conventional robust control approaches like feed-forward H2 and feedback H�‡ methods to show the capability of recurrent neural networks in a robust control paradigm.