Warm Forming Characteristics of AA7075: Microstructure Interaction Mechanisms and Constitutive Models

TL;DR

This paper studies how AA7075 aluminum alloy behaves during warm forming, using experiments and models to predict its flow stress accurately.

Contribution

A novel hybrid machine-learning model combining HHO and LSTM is proposed for predicting flow stress in AA7075 warm forming.

Findings

Dynamic recovery is the main softening mechanism during warm deformation of AA7075.

Flow stress decreases with increasing temperature and increases with increasing strain rate.

The HHO-LSTM model outperforms traditional models in predicting flow behavior.

Abstract

The AA7075 holds significant importance in the aerospace field. Understanding its microstructure evolution and constitutive relationships during warm deformation is crucial for optimizing forming processes. To this end, isothermal compression experiments were conducted at different temperatures and strain rates to analyze their flow stress behavior. The microstructure evolution was characterized using electron backscatter diffraction (EBSD) and transmission electron microscopy (TEM). Microstructural analysis confirmed that dynamic recovery constitutes the predominant softening mechanism under warm forming conditions. The results indicate that flow stress is highly sensitive to deformation parameters, decreasing with increasing temperature and rising with increasing strain rate. To accurately describe the flow behavior, two distinct constitutive models were formulated: (1) a phenomenological Hensel–Spittel–Garofalo (HSG) model; (2) a novel hybrid machine-learning model that innovatively integrates the Harris Hawks Optimization (HHO) algorithm with an LSTM model. Both constitutive models demonstrate reasonable predictive accuracy. In comparison, the HHO-LSTM model demonstrated a superior ability to capture complex nonlinear relationships, achieving highly precise predictions of flow stress across the full range of deformation conditions tested in this work. The hybrid machine-learning model proposed in this study provides a highly accurate method for describing and predicting the flow behavior of the AA7075 during warm forming, offering a powerful predictive tool for engineering applications.

Figures