基于人工神经网络的锻造性能预报的研究  

Prediction of Forgability Based on Artificial Neural Network

作　　者：郭斌[1] 戴护民[1] 吴胜新 

机构地区：[1]哈尔滨工业大学材料工程系

出　　处：《哈尔滨工业大学学报》1998年第5期16-19,共4页Journal of Harbin Institute of Technology

基　　金：哈工大校基金

摘　　要：应用Ｇｌｅｅｂｌｅ－１５００热模拟试验机研究了１Ｃｒ１８Ｎｉ９Ｔｉ的热墩成形过程，得出了流动应力随变形条件的变化规律，测量了空冷后试件的硬度．首次采用人工神经网络对锻造性能进行预报，采用不同的改进ＢＰ算法加速了网络的收敛，得到了较好的网络信息，对１Ｃｒ１８Ｎｉ９Ｔｉ的锻造性能进行了较准确预报．This paper studied the heat forming process of 1Cr18Ni9Ti by using Gleeble1500 Thermo simulation machine. The variation law of stress with forming condition has been established, and the hardness of air cooled workpiece measured. Forgability is predicted for the first time by applying artificial neural network and different improved BP algorithms have accelerated the convergence speed of neural network accurate forgability prediction has been achieved for 1Cr18Ni9Ti.

关 键 词：热力模拟 神经网络 锻造性能 预报 锻件 质量 

分 类 号：TG316[金属学及工艺—金属压力加工]