基于神经网络的6A02铝合金连接板锻造工艺优化.pdfVIP

HotWorkingTechnology2023,Vol.52,No.7

基于神经网络的6A02铝合金连接板

锻造工艺优化

马敬敬，李梦婷

(1.重庆三峡职业学院，重庆404155;2.重庆理工大学材料科学与工程学院，重庆400054)

摘要：采用5×30x10×2四层拓扑结构，输人层为模具预热温度、始锻温度、终锻温度、冷却速度及锻造速度，输

出层为抗拉强度、屈服强度，建立6A02铝合金连接板锻造工艺神经网络优化模型并进行了训练、预测及实际应用。

结果表明：铝合金连接板锻造工艺神经网络优化模型输出的预测抗拉强度相对误差在1.54%~3.29%，平均预测相对

后锻件力学性能提升15%，最优锻造工艺参数为模具预热温度340℃、始锻温度460℃、终锻温度340℃、冷却速度

50℃/s、锻造速度12mm/s。

关键词：神经网络；6A02铝合金；锻造；工艺优化

D0I:10.14158/j.cnki.1001-3814.

中图分类号：TG306文献标识码：A文章编号：1001-3814(2023)07-0106-04

ForgingProcessOptimizationof6A02AluminumAlloyConnectingPlate

BasedonNeuralNetwork

MAJingjing,LIMengting?

(1.ChongqingThreeGorgesPolytechnicCollege,Chongqing404155,China;2.SchoolofMaterialsScienceandEngineering,

ChongqingUniversityofTechnology,Chongqing400054,China)

Abstract:Takingthediepreheatingtemperature,initialforgingtemperature,finalforgingtemperature,coolingrateand

forgingspeedastheinputlayer,andthetensilestrengthandyieldstrengthastheoutputlayer,theneuralnetworkoptimization

modelforforgingprocessof6A02aluminumalloyconnectingplatewasestablishedbyusing5x30x10x2four-layertopology

structure,andthemodelwastrained,predictedandapplied.Theresultsshowthattheneuralnetworkoptimizationmodelof

theforgingprocessofaluminumalloyconnectingplateshasarelativeerroroftensilestrengthpredictionbetween1.54%and

3.29%,andanaveragepredictionerroris2.03%;thepredictedrelativeerrorofoutputyieldstrengthisbetween0.88%and

2.98%,theaveragepredictedrelativeerroris1.43%,andtheoverallrelativeerrorissmall;themechanicalpropertiesofthe

forgings