Self-adaptive vibration control of smart composite beams using recurrent neural architecture

Abstract

A self-adapting vibration control system is developed for damping augmentation in smart composite beams. The conventional vibration control approaches are limited by the requirement of an explicit and often accurate identification of the system dynamics and subsequent offline design of an optimal controller. In the present study a self-adapting vibration control system is developed. A hybrid system comprised of a dynamic diagonal recurrent neural network (DRNN) and an adaptable feed forward neural network is used to control the beam vibrations. Sensing and actuation are achieved using piezoelectric sensors and actuators. A finite element model based on a higher-order shear deformation theory is used to simulate the vibration response of laminated composite beams with integrated piezoelectric sensors and actuators. The dynamic effects of mass and stiffness of the piezoelectric patches are considered in the model. The performance of the DRNN controller is verified for arbitrary initial conditions and loadings. A robustness study including the effects of tip mass, structural parameter variation and partial loss of sensor output is performed. The performance with partial failure of control actuation is also examined. It is seen that the robustness and control capabilities of the hybrid control system are excellent.