Deep learning based reduced order modeling of seismogram-type acceleration time series model: Part - III

Abstract

The optimal deep learning optimizer can vary depending on the features of the solution of the forward model being trained. The other parameter that can acquire significance in LSTM based encoder decoder frameworks is the number of stacked LSTM layers per encoder and decoder. In this work, we zero-in on the optimal optimizer and number of layers for the reconstruction of the acceleration time series solution of the spring slider damper idealization of a friction model.