Physics-Informed Multi-LSTM Networks for Metamodeling of Nonlinear Structures
By: Tashi Rai
Department: Engineering
Faculty Advisor: Dr. Cheng Chen
Non-linear behavior of structures pose uncertainty and unpredictable response of structures during earthquake. With the use of deep neural network and its powerful machine learning capability, we can learn the dynamics of the structural responses and get closer to predicting its behavior. It is accomplished by embedding the law of motion equation in the code to train model so that the noises in the data are minimized. Two physics-informed deep long short-term memory (LSTM) networks are created and simulated to capture the non-linearity of structures.