Predictive modelling of creep age forming parameters using artificial neural networks
Abstract Creep age forming (CAF) is an advanced shaping technique for high-strength aluminum alloys used in the aerospace and automotive industries. This process combines creep deformation and age hardening to shape complex parts while enhancing their strength. Optimizing CAF parameters, however, re...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-07-01
|
| Series: | Discover Applied Sciences |
| Online Access: | https://doi.org/10.1007/s42452-025-07518-9 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Creep age forming (CAF) is an advanced shaping technique for high-strength aluminum alloys used in the aerospace and automotive industries. This process combines creep deformation and age hardening to shape complex parts while enhancing their strength. Optimizing CAF parameters, however, remains challenging due to the complex interplay among material properties, creep deformation, and ageing mechanisms. This study explores the use of artificial neural networks (ANNs) as a predictive tool for CAF parameter optimization, focusing on AA7050-T6 aluminum alloy. The ANN model was trained using experimental data from 80 samples, incorporating stress (75–175 MPa), ageing time (0–8 h), and creep strain as input parameters to predict precipitate radius and yield strength. The developed ANN demonstrated exceptional accuracy, achieving R2 values of 0.99 for both outputs with remarkably low error metrics: RMSE of 0.237 nm for precipitate radius and 1.405 MPa for yield strength, and MAPE values below 1% for all predictions. This data-driven approach effectively captured complex, non-linear interactions between process parameters and material properties, significantly outperforming traditional constitutive models, particularly at higher stress levels (> 150 MPa) and prolonged ageing durations (> 4 h). The ANN accurately predicted stress-accelerated precipitate coarsening behavior, with experimental validation showing precipitate growth from 15 to 52 nm under creep conditions versus 47 nm under standard ageing. Yield strength predictions ranged from 456 to 573 MPa with exceptional precision across all stress and time combinations. The superior performance over constitutive models was most pronounced at extended ageing times, where the ANN maintained accuracy while traditional models showed increasing divergence from experimental data. The ANN’s ability to accurately predict key CAF parameters demonstrates potential for reducing experimental effort by up to 70%, improving process efficiency, and enabling rapid optimization of ageing conditions for target mechanical properties. This methodology offers a scalable, accessible solution for industrial CAF applications, advancing the development of lightweight, high-performance materials for demanding applications in the aerospace and automotive sectors. |
|---|---|
| ISSN: | 3004-9261 |