基于BP神经网络的Ti40钛合金流变性能的预测与评估

Prediction and Evaluation of Rheological Properties of Ti40 Alloy Based on BP Neural Network

在变形温度900-1100℃，应变速率O．01～10S^1的条件下，运用Gleeble－1500热模拟机对Ti40钛合金进行等温热压缩试验，分析了Ti40钛合金的热流变行为，运用训练组数据计算得到ANN预测模型，通过相关性系数和相对误差等一系列统计标准预测并评估了ANN模型的预测能力。结果表明：基于ANN建立的本构关系模型的预测值与实验值吻合良好．并且此ANN模型可用于描述Ti40钛合金在热加工时各参数之间的高度非线性关系，以及能用于非实验条件下Ti40钛合金流变行为的预测，这为Ti40钛合金本构关系模型的建立提供更加准确有效的方法。

Isothermal compression deformation tests were conducted for Ti40 alloy by Gleeble-1500 thermal simulator. According to the obtained experimental data （deformation temperatures range from 900℃ to 1100 ℃ and strain rates range from 0.01 s^-1 to 10 s^-1）, the thermal rheological behavior of the alloy was analyzed. An artificial neural network （ANN） was trained with standard back-propagation learning algorithm, and a series of standards （the correlation coefficient and relative error） were used to evaluate the predictive ability of the ANN model. The results show that the feed-forward hack-propagation ANN model can accurately track the experimental data, which can describe the complicated nonlinear relationship of thermo- dynamical parameters well and can be used for theological behavior prediction of the titanium in the non-experimental conditions. Therefore, it provides more convenient and more effective ways to establish the model of constitutive relationship for Ti40 alloys.