铸造2025，Vol.74Issue(10)：1336-1343,8.
基于机器学习的挤压铸造A356合金右悬置托臂全域力学性能预测
Prediction of Global Mechanical Properties of Squeeze-Cast A356 Alloy Right Suspension Bracket Using Machine Learning
摘要
Abstract
This paper proposes a machine learning-based method to predict the global mechanical properties of squeeze-cast A356 alloy components.A Back Propagation Neural Network(BPNN)model was developed with 3 input nodes,2 output nodes,and an 11×5 hidden layer architecture.Process data were derived from temperature field simulations in ProCast software,and combined with microstructure feature identification and local mechanical property predictions to construct the BPNN dataset.The model was trained and validated using Mini-batch Gradient Descent(MBGD)and an extrapolation strategy,establishing a mapping between processing parameters,microstructure,and properties.This enabled prediction of the tensile strength and elongation distribution across a right suspension bracket,with results validated through tensile testing.Analysis of the model's weights and biases led to a polynomial regression model that elucidated the parameter optimization process,improving interpretability and generalization ability.The findings showed that this method offers an efficient,cost-effective approach to predict global mechanical properties of squeeze-cast A356 alloy components,supporting the structural optimization design.关键词
A356合金/挤压铸造/机器学习/反向传播神经网络Key words
A356 alloy/squeeze casting/machine learning/back propagation neural network分类
金属材料引用本文复制引用
钟皓南,马万里,白文辉,黄智,杨鹏,赵海东..基于机器学习的挤压铸造A356合金右悬置托臂全域力学性能预测[J].铸造,2025,74(10):1336-1343,8.基金项目
北京市自然科学基金-小米创新联合基金(L223001) （L223001）
广东省重点领域研发计划资助项目(2020B010186002). （2020B010186002）
