Abstract: The each technology parameters in range of carbon 0.58% -0.83%, water ratio in secondary cooling zone 0.82-2.86 L/kg, casting speed 1. 9 〜3.0 m/min and superheating of liquid steel 17 ~ 60 ℃： on center carbon segregation of 140 mm x 140 mm and 150 mm x 150 mm billet has been analyzed at given mould electromagnetic stirring (230 ~ 245 A) using multivariate linear regression analysis, and the predicted calculation was carried out using Back Propagation (BP) neural networks. Hie results by multivariate regression analysis showed that the carbon content in steel was remarkable factor to influence the center carbon segregation, next were casting speed and water; and with carbon content in steel, casting speed and superheating of liquid steel increasing, the center carbon segregation in billet increased and with water ratio in secondary cooling zone increasing, the center carbon segregation decreased.

摘要： 采用多元线性回归方法分析了在给定结晶器电磁搅拌条件下(230～245A),碳含量0.58%～0.83%,二冷比水量0.82～2.86L/kg,拉速1.90～3.0m/min,过热度17～60℃范围内,各参数对140mm×140mm和150mm×150mm小方坯中心碳偏析的影响,并采用BP神经网络进行预测计算。多元回归分析结果表明,影响中心碳偏析的显著因素是钢中碳含量,其次是连铸拉速和二冷比水量;随钢中碳含量、拉速和过热度增加,铸坯中心碳偏析增大,随二冷比水量增加,中心碳偏析减小。

关键词: 高碳钢,连铸小方坯, 中心碳偏析, 多元线性回归, BP神经网络