应用人工神经网络模型构造TC21合金加工图  

Processing map of TC21 alloy established on artificial neural network model

作　　者：于雪梅 周舸 刘波 刘丽荣[1] YU Xue-mei;ZHOU Ge;LIU Bo;LIU Li-rong(School of Materials Science and Engineering,Shenyang University of Technology,Shenyang 110870,China;Tieling Country High School,Tieling 112002,China)

机构地区：[1]沈阳工业大学材料科学与工程学院,辽宁沈阳110870 [2]铁岭县高级中学,辽宁铁岭112002

出　　处：《塑性工程学报》2018年第6期250-256,共7页Journal of Plasticity Engineering

基　　金：辽宁省教育厅科学研究一般项目(LQGD2017024)

摘　　要：采用Gleeble-3500热压缩模拟试验机对TC21合金进行了不同温度、不同应变速率、不同变形量下的单道次热压缩模拟实验。以实验数据为基础,建立了以热变形温度、应变速率、变形量为输入变量,流变应力为输出变量的3层BP人工神经网络模型,较为准确的预测了该合金不同热成形工艺区间内流变应力与热变形工艺参数之间的定量关系,平均预测误差为0. 264%。应用动态DMM热加工图理论,绘制了该合金基于Prasad失稳判据下热加工图。结合TC21合金热变形过程中的微观组织演变,对其热成形性能及变形机理进行了分析,结果表明:在温度为820~890℃,应变速率为5×10-4~4×10-3s-1范围内该合金功率耗散率为62%,此时该合金在变形过程中发生了动态再结晶及α→β相转变现象,具有良好的热加工成形性能。Single-pass thermal compression simulation experiment of TC21alloy was carried out on Gleeble -3500 thermal compression simulating tester at different temperatures,strain rates and strains.Based on the experimental data,a three-layer BP artificial neural network model was established.In the network model,the input parameters were deformation temperatures,strain rate and strain,the output parameter was flow stress.The quantitative relationship between flow stress and thermal deformation process parameters in different thermal deformation processes intervals was accurately predicted by the model,and the average prediction error was 0.264%.Thermal processing map of the alloy was calculated based on the Prasad instability criterion under the theory of dynamic DMM processing map.The thermal forming properties and deformation mechanism of TC21 alloy were analyzed based on microstructure evolution in thermal deformation process of the alloy.The results show that the dissipation rate of the alloy is 62%at 820-890℃ and 5×10^-4~4×10^-3s^-1.By this time,the dynamic recrystallization and α→β transformation occur in the alloy,which contributes to excellent thermal forming properties.

关 键 词：TC21合金 加工图 BP人工神经网络 热变形行为 

分 类 号：TG156[金属学及工艺—热处理]