Structural Damage Control Using Neuro-Ga Algorithm

In this paper a semi-active structural damage control strategy using MR dampers and Neural Networks is presented. A multilayer feed-forward neural network has been designed. The input layer is relative displacement of stories and the output layer is the voltage needed for MR damper. The neural network is learned to predict the voltage needed for MR damper that can minimize the Park & Ang damage index of structure. Genetic algorithm has been used to learn the neural network. The Park & Ang damage index of the structure has been used as the fittnees function of the genetic algorithm. To evaluate the structural control system a nonlinear 3 story benchmark building has been selected. The results show the the proposed structural control system can effectively reduce the Park & Ang damage index of the structure.