High Cycle S-N curve prediction for Al 7075-T6 alloy using Recurrent Neural Networks (RNNs)

TL;DR

The paper addresses the costly acquisition of high-cycle fatigue data for Aluminum 7075-T6 and proposes a TR-LSTM framework that transfers a model trained on axial rotating-bending fatigue to predict reversed torsion S-N curves. It demonstrates that the TR-LSTM achieves a torsional RMSE of 0.53 MPa, vastly outperforming a non-transfer LSTM (71.39 MPa), while the axial source model attains 27.63 MPa. This approach can substantially reduce fatigue-characterization costs and help prioritize tests under time and budget constraints, though its generalization to other alloys is limited by data availability and may benefit from exploring Markovian models in future work.

Abstract

Aluminum is a widely used alloy, which is susceptible to fatigue failure. Characterizing fatigue performance for materials is extremely time and cost demanding, especially for high cycle data. To help mitigate this, a transfer learning based framework has been developed using Long short-term memory networks (LSTMs) in which a source LSTM model is trained based on pure axial fatigue data for Aluminum 7075-T6 alloy which is then transferred to predict high cycle torsional S-N curves. The framework was able to accurately predict Al torsional S-N curves for a much higher cycle range. It is the belief that this framework will help to drastically mitigate the cost of gathering fatigue characteristics for different materials and help prioritize tests with better cost and time constraints.

Figures (3)

• Figure 1: Schematic of TR-LSTM for S-N curve: (a) the framework of transfer learning in this research; (b) LSTM cell; (c) the architecture of LSTM model. WEI2022107050
• Figure 2: Prediction results for Axial data
• Figure 3: Prediction results for Torsional Data