摘要
采用RBF神经网络对204组X70管线钢生产数据进行训练,建立了管线钢成分与力学性能的预测模型,经检验该模型预报精度高,网络预报值与实际值较吻合。利用此模型预报了C、Mn、Mo、Nb、V、Ti等元素含量对管线钢性能的影响规律,并在此基础上确定了X80管线钢的成分范围。对试制生产的X80管线钢进行组织性能检测,结果表明,X80钢的显微组织主要由针状铁素体和粒状贝氏体组成,晶粒细小,力学性能指标达到X80管线钢应用要求。
A radial basis function (RBF) artificial neural network mapping model for composition and mechanical properties of pipeline steels was established based on using nearest neighbor clustering of 204 sets of actual production data. The trained network model exhibits higher accuracy in the prediction, and the prediction values and measured values are coincident very well. Using this model, a test of forecasting on the effects of contents of the alloying elements of C, Mn, Mo, Nb, V, Ti on mechanical properties was conducted, and composition range of X80 pipeline steel was identified. The microstrueture and mechanical properties of X80 pipeline steel were tested. The results show that the mierostructure of X80 pipeline steel is mainly composed of acieular ferrite and granular bainite, and the grain size is fine and the mechanical properties of the X80 pipeline steel fulfill the specifications for engineering application.
出处
《材料热处理学报》
EI
CAS
CSCD
北大核心
2012年第12期152-157,共6页
Transactions of Materials and Heat Treatment
基金
湖南省重大科技专项
关键词
RBF神经网络
成分设计
X80管线钢
组织性能
RBF neural network
composition design
X80 pipeline steel
microstrueture and property