OPTIMUM DESIGN OF VIBRATING BARRIERS USING ARTIFICIAL NEURAL NETWORK

Vibrating Barrier (or ViBa) is a newly proposed passive, non-invasive technique for seismic protection of structures. It is described as a pre-tuned spring-mass-damper device that is installed underground near the host structure. Its main advantage is its ability to absorb seismic energy, reducing the demands on the structure without ever touching it. In this paper a new approach for ViBa design is proposed by utilizing the power of Artificial Neural Network (ANN). In this regard several sequential neural networks are tested on a case study structure, the step pyramid in Giza, Egypt, in order to optimize the ViBa parameters. The predicted ViBa parameters were tested using a numerical model for the step pyramid to measure the accuracy of the proposed architectures. The peak step pyramid acceleration is reduced by 46% using ANN approach compared to 30% using the traditional stochastic approach. Additional results using inelastic finite element analysis also confirmed the advantages gained from using the Viba.