基于神经元网络的热暴露对TC4钛合金拉伸性能影响预测

Prediction of Effect of Thermal Exposure on Tensile Properties of TC4 Titanium Alloy Based on Neural Network

对TC4钛合金热处理态、200、400和500℃热暴露不同时间的拉伸性能进行了研究,并利用BP人工神经元网络方法建立了不同温度与时间热暴露下试样拉伸性能的预测模型。结果表明:大多数热暴露与拉伸测试条件下合金的强度与塑性性能并未发生严重的恶化,热暴露前后试样拉伸塑性的差值大多在±7.5%左右波动,断面收缩率最多降低了约13%。对所建立的BP人工神经网络模型预测精度的分析表明,当隐含层神经元个数为11时,该模型的预测效果最佳。该模型能够很好地预测TC4合金不同热暴露条件下拉伸性能的变化。

The tensile properties of TC4 titanium alloy with heat treatment at 200 ℃, 400 ℃ and 500 ℃ for different time were studied. The BP artificial neural network method was used to establish the prediction model of the tensile properties of the samples after heat exposure under different temperature and time. The results show that, the strength and plastic properties of the alloy do not deteriorate seriously under the conditions of thermal exposure and tensile test, and the difference of tensile plasticity before and after thermal exposure is mostly about ±7.5%. The shrinkage of the section is reduced by up to about 13%. The prediction accuracy of the BP artificial neural network model shows that, the optimal prediction accuracy is obtained when the nodal number of the hidden layer is no less than 11. Therefore, the established model can predict tensile properties of TC4 alloy very well under different thermal exposure conditions.