摘要
结晶器是钢水凝固成型的核心设备,其内部的传热和摩擦直接决定铸坯的表面裂纹和漏钢等各类异常,是实现高效连铸的关键因素.基于功率法检测到的板坯结晶器摩擦力实测数据,对摩擦力的异常预报方法进行了研究.建立了以BP人工神经网络为基础的异常预报模型,并开发出相应软件.对应现场的异常记录,离线预报结果表明:软件能够对漏钢、水口断裂及液位波动等各类异常进行预报,并具有一定的预报提前量,证明了方法的可行性,并显示出极大的应用潜力.
The mould is the core instrument for interior heat transfer and friction of mould are prim close ary cooling and slab forming of liquid steel. The ly related to the surface defects and breakout, which is very important for effectively continuous casting. Based on the online measured data of mould friction by power-method in slab continuous casting, the method of prediction for abnormalities in continuous casting was studied. A model has been built using BP neural networks, and the software of prediction is also developed. According to the abnormal records of steel plant, the results of simulating prediction show that the acute mould level fluctuations and software can predict breakout, other abnormalities, and can submerged entry nozzle broken, predict earlier than temperature system before some abnormalities happen. The method shows that it is feasible to predict the abnormalities in slab continuous casting, and great potential in application of this method is demonstrated.
出处
《大连理工大学学报》
EI
CAS
CSCD
北大核心
2008年第4期514-518,共5页
Journal of Dalian University of Technology
基金
国家自然科学基金资助项目(50174010)
教育部科学技术研究重点资助项目(03051)
关键词
连铸
结晶器
异常
神经元网络
continuous casting
mould
abnormality
neural network