3J33B马氏体时效钢时效工艺-时效硬度的人工神经元预报  

Prediction of aging process-hardness of maraging steel based on BP neural network

作　　者：曹勇[1] 李殿生 朱景川[1] 刘勇[1] 来忠红[1] 

机构地区：[1]哈尔滨工业大学材料科学与工程学院,黑龙江哈尔滨150001 [2]东北特钢集团,辽宁大连116031

出　　处：《金属热处理》2011年第11期107-110,共4页Heat Treatment of Metals

基　　金：国家自然科学基金委员会"重点学术期刊专项基金"(50131010)

摘　　要：对不同时效处理的3J33B马氏体时效钢进行硬度测试,获得了时效工艺(温度、时间)、硬度参数数据。利用BP人工神经网络建立起其关系网络模型。结果表明,所建立的网络可以很好地反映出材料的时效工艺-时效硬度之间的关系,网络模型可以用来预测不同时效条件下3J33B马氏体时效钢的时效硬度,并且利用粒子群优化,对3J33B马氏体时效钢的时效工艺进行优化,对实际生产具有有效的指导作用。Parameters of processing（aging temperature,time） and aged hardness of maraging steel were obtained through mechanical properties examination,and their relationship network model was built by BP artificial neural network.The results show that the built model can reflect the relationships between the processing and the aged hardness very well and is certainly accurate.It can be used for predicting the properties of 3J33B steel under different aging process.Meanwhile,the optimized aging temperature and time can be obtained with particle swarm optimization.The model can serve as a guide for the aging treatment of maraging steel.

关 键 词：马氏体时效钢 人工神经网络 粒子群优化 

分 类 号：TP389.1[自动化与计算机技术—计算机系统结构]