The breakout and other abnormalities in the mould have a critical influence on strand surface and casting productivity in continuous casting. Prediction of these mould abnormalities is an essential prerequisite for producing a good quality product with minimal disruption and defects. It is also important for optimization of casing process and operating safety. The mould friction (MDF) between strand and mould is one of the most important parameters that can be used to describe frictional behavior and mechanical interaction between the strand and the mould. Monitoring MDF online can contribute to evaluation of the powder lubrication, optimization of casting variables, prediction to breakout and other abnormities. In this work, based on the measurement data of mould friction on two strands slab caster in a steel plant, the prediction method for MDF abnormalities has been investigated by using artificial neural network models in combination with two auxiliary algorithms, one for ramp and the other for pulse variation of MDF. A set of software to predict the MDF abnormalities of continuous casting has been developed. The results of simulating prediction for online measurement MDF data are found to be basically consistent with those collected from the abnormal records of steel plant.
[1]
D. Dunne,et al.
Artificial Neural Networks for Modelling of the Impact Toughness of Steel
,
2004
.
[2]
S. J Gartland,et al.
Neural network methods for corrosion data reduction
,
1999
.
[3]
A. S. Normanton,et al.
Mould thermal monitoring: a window on the mould
,
2004
.
[4]
Brian G. Thomas,et al.
Investigation of strand surface defects using mould instrumentation and modelling
,
1994
.
[5]
A. S. Normanton,et al.
Developments in online surface and internal quality forecasting of continuously cast semis
,
2004
.
[6]
J. Watzinger,et al.
MoldExpert – operational experience and future development
,
2005
.
[7]
M. Yao,et al.
On line measuring method for mould friction in continuous casting
,
1996
.