Vibration control of smart composite laminated spherical shell using neural network

This study presents a neural network approach for the identification and control of a smart composite laminated spherical shell. The spherical shell is in the form of a layered composite shell having a sensor and an actuator layer. The vibratory response of the shell is modeled using FEM. A degenerate shell element is implemented to model composite laminated spherical shell. Modeling is based on the first-order shear deformation theory and linear piezoelectricity theory. The mode superposition method has been used to transform the coupled finite element equations of motion in the physical coordinates into a set of reduced uncoupled equations in the modal coordinates. The reduced uncoupled equations are transformed into discrete state space form. An identifier neural network has been trained using the results of the FEM program to predict the future response of the structure from the immediate history of the system's response. Then a controller neural network has been trained with the aid of the identifier neural network so that the overall behavior of the controlled system can be described by a prescribed reference model. Numerical results have been presented for the vibratory response of the laminated composite spherical shell. The controlled response of the shell is found to exactly follow the reference model.