基于BP神经网络的GH5188高温合金本构模型

采用Gleeble-3500热模拟试验机对GH5188高温合金试样进行热压缩试验,研究其在应变速率为0.001~0.1s^-1和变形温度在1000~1150℃时的热变形行为;建立了基于BP神经网络的本构模型,并验证了所建本构模型的可靠性,最后基于误差计算分析了BP神经网络本构模型的精度。结果表明,温度和应变速率对GH5188合金流变应力的影响明显,随着压缩温度升高和应变速率降低,GH5188合金流变应力明显减小。经定量误差计算分析,BP神经网络本构模型应力预测偏差值在10%以内的数据点占97.92%,BP神经网络模型能准确地预测GH5188高温合金的高温流变应力。

Constitutive Model of GH5188 Superalloy Based on BP Neural Network

The thermal deformation behavior of GH5188 superalloy was investigated by the hot compression on the Gleeble-3500 hot simulation compression testing machine with the strain rate range of 0.001~0.1 s^-1and the deformation temperature range of 1000~1150℃.A constitutive model based on BP neural network was established,and the reliability of the constructed constitutive model was verified.Finally,the accuracy of BP neural network constitutive model based on error calculation was analyzed.The results indicate that the effect of temperature and strain rate on flow stress of GH5188 superalloy is very significant,and the flow stress of GH5188 superalloy is decreased greatly with the increasing in compression temperature or decreasing in the strain rate.After quantitative error calculation and analysis,date points where stress prediction deviation value of BP neural network constitutive model is within 10%account for 97.92%.Therefore,the BP neural network model can accurately predict the high temperature flow stress of GH5188 superalloy.