Indentation Reverse Algorithm of Mechanical Response for Elastoplastic Coatings Based on LSTM Deep Learning.

Area of Science:

• Materials Science
• Computational Materials Science
• Machine Learning Applications

Background:

• Nanoindentation experiments yield load-penetration depth (P-h) curves for metallic coatings.
• Extracting stress-strain response and elastoplastic properties directly from P-h curves is challenging.
• Conventional finite element (FE) methods for this analysis are time-consuming and lack generality.

Purpose of the Study:

• To develop an efficient method for determining elastoplastic properties from nanoindentation data.
• To overcome the limitations of traditional FE simulations for P-h curve analysis.
• To establish a deep learning model capable of mapping P-h curves to stress-strain responses.

Main Methods:

• A long short-term memory (LSTM) neural network was developed to learn the time series of P-h curves.
• 1000 sets of FE-generated indentation data (P-h curves and stress-strain curves) were used for training and validation.
• Hyperparameter optimization of the LSTM network was performed by comparing loss curves.

Main Results:

• The LSTM neural network accurately predicted the relationship between P-h curves and stress-strain responses for metallic coating materials.
• The predicted relationship closely matched the power-law equation.
• The deep learning approach demonstrated significantly higher efficiency compared to FE analysis.

Conclusions:

• Deep learning based on LSTM offers an advantageous and efficient method for interpreting elastoplastic behaviors from indentation measurements.
• The established LSTM model achieves a prediction accuracy of up to 97%, meeting practical engineering requirements.
• This method provides a reliable and rapid alternative for material property characterization.