Deep neural network in the structural analyses

• Grant number: 19/04747-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2019
• End date: April 30, 2020
• Field of knowledge: Engineering - Mechanical Engineering - Mechanics of Solids
• Principal Investigator: Larissa Driemeier
• Grantee: Gabriel Lopes Rodrigues
• Host Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

Understanding structural analysis in the context of artificial intelligence is the major challenge of the proposed project. In this work, the prediction of the response using artificial intelligence of a simple structural problem of lattice bars will be investigated. Although there is a large number of mathematical methods available to describe the linear behavior of a bar structure, the goal is to understand deep neural networks and investigate state-of-the-art learning techniques such as ReLU, dropout, mini-batch, etc. In this way, the work aims to analyze the applicability, comprehensiveness and accuracy of the deep neural networks in a simple structural analysis. The finite element numerical analysis will be used to generate network training and validation data.