(1. 上海电力学院 数理学院,上海 201399;
2. 上海大学 计算机工程与科学学院,上海 200444)
摘 要: 根据一系列 Al基非晶合金薄带实测数据集,应用粒子群优化支持向量回归方法(PSO-SVR),建立一个通过相关表征参数来预测Al基非晶合金晶化温度(Tx)的模型。利用该模型对不同类型铝基非晶合金的晶化温度(Tx)进行建模和预测研究,并与反向传播神经网络(BPNN)预测方法进行比较。结果表明:基于留一交叉验证法 (LOOCV)的PSO-SVR模型预测的晶化温度误差要比BPNN模型预测的小得多,这说明模型中所采用的特征参数能很好地描述该系列Al基非晶合金的晶化行为和热稳定性。
关键字: Al基非晶合金;晶化温度;支持向量回归;粒子群优化
(1. School of Mathematics and Physics, Shanghai University of Electric Power, Shanghai 201399, China;
2. School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China)
Abstract:According to the experimental data of Al-based amorphous alloys, a model to predict the crystallization temperature Tx of Al-based amorphous alloys by using particle swarm optimization combined with support vector regression (PSO-SVR) was established. Based on this model, crystallization temperature Tx can be predicted, and then compared with the method of back-propagation neural network (BPNN). The results show that the prediction error is smaller by using PSO-SVR. This means that the crystallization behavior and thermal stability of Al-based amorphous alloys can be well described by the parameters used in PSO-SVR model. Moreover, the PSO-SVR model could provide an important theoretical and practical guidance to the research on Al-based amorphous alloys.
Key words: Al-based amorphous alloy; crystallization temperature; support vector regression; particle swarm optimization