Al-Cu-Mg-Ag合金强度性能的支持向量回归预测

Strength Prediction for Al-Cu-Mg-Ag Alloy Based on Support Vector Regression

为了研究不同时效工艺下Al-Cu-Mg-Ag合金强度性能,根据实测数据集,应用基于粒子群算法(PSO)寻优的支持向量回归(SVR)方法,建立了SVR预测模型。模型以Al-Cu-Mg-Ag合金时效温度与时效时间为输入,合金的抗拉强度、屈服强度为输出。经过与BP神经网络模型进行比较,结果表明:对于相同的训练样本和检验样本,支持向量回归模型比BP神经网络模型具有更高的预测精度。

To explore the strength properties of Al-Cu-Mg-Ag alloy at different aging temperature and aging time, the support vector regression （SVR） approach combined with particle swarm optimization （PSO） algorithm was proposed to construct a SVR model based on experimental data. In the modeling process, the aging temperature and aging time were employed as input parameters, the tensile strength and yield strength acted as outputs. By comparison with BP neural network, it was found that the prediction accuracy of the established SVR model was higher than that of BPNN model by applying identical training and test samples. This investigation would provide a theoretical foundation for further study on the effect of aging condition on mechanical property, and the optimal design of the aging process for fabricating Al-Cu-Mg-Ag alloy.