Center for Energy and Environmental Sustainability

2019-2024 National Science Foundation (NSF) #1914692, Amount: $5,000,000 (5 years): CO-PI

Sub-project 2: Offshore Wind Energy

Overview: The research is to enable the structural health monitoring (SHM) using Long-Term-Short-Memory (LSTM) neural network, a Recurrent Neural Network (RNN) known as a Deep Learning (DL) technique. The research is focusing on the damage detection of the turbine blades based on the accelerometer signal of the turbine and the wind power prediction based on the historic wind speed data.  The research objectives of my subproject is to (1) establish the framework of structural health monitoring using LSTM networks (2) enable to perform the SHM using the DL technique, (3) find the prediction errors of the SHM and  the relationship between the prediction errors, number of the data sets, and desired prediction length. (4) to explore further enhancement methods to minimizing the SHM prediction errors.

BasicLSTM

Basic LSTM

BiLSTM

Bi-directional LSTM