摘要
BP神经网络预测模型可用于相关材料组织、性能与相关参数对应关系的预测,但目前还未见用于热浸镀铝层(表面层和金属间化合物层)厚度的预测。利用溶剂法对Q235钢热浸镀铝,以正交试验法分析了浸镀温度、时间、提升速度、浸镀液中硅含量4种因素对铝镀层厚度的影响,建立了相关的神经网络预测模型。结果表明:4种因素对铝表面层厚度影响的大小顺序为提升速度>硅含量>浸镀温度>浸镀时间;对金属间化合物层厚度影响则为浸镀温度>浸镀时间>硅含量>提升速度;利用正交试验数据对建立的BP神经网络预测模型进行训练后,对镀铝层厚度的预测结果与试验结果相符。
Hot-dipped aluminum coating was prepared on the surface of Q235 by a solvent method.The effects of dipping temperature,dipping time,extractive velocity and content of Si in the plating bath on the thickness of the surface layer and intermetallic layer of the Al coating were investigated by orthogonal tests,and a model was established to predict the thicknesses of the two layers based on neural networks.Results indicate that various factors affecting the surface layer thickness of the hot-dipped Al coating are ranked as extractive velocity>content of Si in plating bath>dipping temperature>dipping time;and those factors affecting the thickness of the intermetallic layer are ranked as dipping temperature>dipping time>content of Si in the bath>extractive velocity.After the established neural networks model is trained based on orthogonal test data,it can be well used to predict the thickness of the hot-dipped Al coating on steel substrate,and the predicted results are in good agreement with the experimental ones.
出处
《材料保护》
CAS
CSCD
北大核心
2012年第8期4-6,70,共3页
Materials Protection
基金
国家自然科学基金(51005140)
山东省自然科学基金(ZR2010EQ037)资助
关键词
预测模型
热浸镀铝
铝层厚度
神经网络
正交试验
prediction model
hot-dipped Al coating
thickness of Al coating
neural networks
orthogonal test
