Review of Neural Network Modeling of Shape Memory Alloys

Area of Science:

• Materials Science
• Artificial Intelligence
• Mechanical Engineering

Background:

• Shape memory alloys (SMAs) are smart materials with unique shape memory effects, widely used in sensors, actuators, robotics, aerospace, civil engineering, and medicine.
• The nonlinear behavior of SMAs complicates traditional modeling methods like the finite element method, leading to increased computation time.
• Developing new AI-based approaches is crucial for efficient and accurate SMA modeling.

Purpose of the Study:

• To review the importance and application of AI in modeling shape memory alloys (SMAs).
• To highlight the deep connection between artificial neural networks (ANNs) and SMAs in diverse fields.
• To analyze various ANN types used for modeling SMA properties in different configurations.

Main Methods:

• Review of existing literature on AI, machine learning, and deep learning applied to SMAs.
• Summarization of general characteristics of ANNs and SMAs.
• Analysis of specific ANN architectures and learning techniques for modeling SMA properties.

Main Results:

• AI, particularly ANNs, has shown success in efficiently modeling SMA features.
• ANNs are effectively applied to characterize SMAs in various forms like wires, springs, magnetic materials, and reinforced concrete beams.
• The review details techniques for NN architectures and learning relevant to SMA modeling.

Conclusions:

• AI provides a promising solution for overcoming the computational challenges in modeling nonlinear SMA behavior.
• ANNs are a powerful tool for characterizing and modeling SMAs, enabling advancements in their applications.
• This review underscores the significant role of AI and ANNs in the future of SMA research and development.