Validation of physics-based response prediction using deep learning

Infrastructures including buildings and bridges show dynamic behaviors due to earthquakes, wind loads, and etc. The excessive dynamic response sometimes leads to structural failure. Therefore, accurate prediction method of the dynamic response for the structures is required to secure the safety the structures. A neural network augmented physics (NNAP) model which is based on deep learning was proposed as an accurate response prediction method for the MDOF structures subjected to dynamic loads. NNAP model complements the conventional numerical model by adding an output of the neural network inside an ordinary differential equation (ODE). The main objective of this project is to validate the deep learning algorithm which predicts the dynamic response of MDOF structures using a hybrid simulation test result. After the test, the proposed method is expected to become more advanced by the collaborative study with Professor Song and Professor Kwon.

Faculty Supervisor:

Oh-Sung Kwon

Student:

Partner:

Seoul National University

Discipline:

Engineering

Sector:

Education

University:

University of Toronto

Program: