Optimized Neural Network Prediction Model of Shape Memory Alloy and Its Application for Structural Vibration Control

Area of Science:

• Materials Science
• Civil Engineering
• Computational Mechanics

Background:

• Traditional mathematical models for shape memory alloys (SMAs) are complex for numerical analysis.
• Artificial neural networks offer a nonlinear modeling approach independent of traditional mathematical models, avoiding inherent errors.
• Developing accurate and efficient models for SMA behavior is crucial for advanced engineering applications.

Purpose of the Study:

• To develop an optimized neural network prediction model for shape memory alloys (SMAs).
• To apply the optimized model for effective structural vibration control.
• To investigate the optimal placement and quantity of SMA wires in a spatial structure for seismic response mitigation.

Main Methods:

• Experimental testing of austenitic SMA wire superelastic properties.
• Training a Backpropagation (BP) neural network using experimental material property data.
• Optimizing the neural network prediction model using a genetic algorithm.
• Employing an improved genetic algorithm to optimize the position and quantity of SMA wires in a three-storey structure.
• Performing dynamic response analysis on the optimized structural configuration.

Main Results:

• The optimized neural network model demonstrates superior agreement with experimental data and higher stability compared to unoptimized models.
• The model accurately captures the rate-dependent superelastic properties of SMAs.
• Optimized placement and quantity of SMA wires significantly inhibit structural seismic responses.
• A configuration with four SMA wires achieved a 44.51% reduction in total structural storey drift, indicating an effective and economical solution.
• Excessive SMA wires do not necessarily improve vibration reduction, highlighting the importance of optimal configuration.

Conclusions:

• The optimized genetic algorithm-BP neural network model provides a high-precision, rate-dependent dynamic prediction for SMAs.
• The developed constitutive model is user-friendly for dynamic simulations of SMA passive control structures.
• Proper configuration of SMA wire quantity and location is essential for effective and economical seismic response reduction in structures.