Danieli & C. Officine Mecchaniche S.p.A. (Buttrio, Italy):

Bianco F., Danieli R & D, e-mail: f.bianco@danieli.it
Dimitrijevic V., Danieli R & D

Danieli Automation (Buttrio, Italy)
Piazza M.

Acciaierie Bertoli Safau (Pozzuolo del Friuli, Italy)
Spadaccini A.

Turco R.

A system for steel temperature prediction during the EAF refning stage and during the vacuum degassing process is described. These systems are based on the predictive capability of a neural network (NN) and they have already been in operation at steel plant as online temperature trackers. After the frst temperature measurement, NNAC temperature tracker starts to continuously predict the molten bath temperature providing a reliable estimation of the bath temperature to the operators. The NNAC tracker supports the operator during the final and most critical stage of the EAF process. The peculiarities of this NN temperature tracker are its simplicity and self-learning capability. The simplicity is referred to the NN topology: its inputs consist of only six continuously stored signals and ten neurons in one hidden layer. The self-learning capability allows the tracker to be automatically retrained on the enlarged learning set by new heats of the most recent campaign, to adapt the tracker output to the process changes over time. The approach described is currently being extended to the ABS DC furnace as well. Similarly, the temperature tracker in vacuum degassing process (NNVD) continuously predicts the bath temperature starting from the last valid LF sample, and has shown good performances. An adhoc strategy is required in order to filter unrealistic input data that may invalidate the temperature estimation. NNVD presence is strategic as it helps the operators to satisfy the stringent superheat temperature target. The implementation of such a tracker is now under consideration for the twin LF-VD dedicated to the DC furnace.

1. Bouhouche, S.; Lahreche, M.; Boucherit, M. S.; Bast, J.: Arab. J. Sci. Eng. 29 (2004) No. 1B, p. 65.
2. Cramb, A. W. (ed.): The Making, Shaping and Treating of Steel — Casting Volume, 11^th ed., The AISE Steel Foundation, Pittsburgh, USA, 2003.
3. Fruehan, R. J. (ed.): The Making, Shaping and Treating of Steel — Steelmaking and Refining Volume, 11^th ed., The AISE Steel Foundation, Pittsburgh, USA, 1998.
4. Kordos, M.; Blachnik, M.; Wieczorek, T.: Lect. Notes in Comput. Sc., ICANN, 6792, (2011), p. 71.
5. Lin, C. T.; George Lee, C. S.: Neural fuzzy systems: A neuro-fuzzy synergism to intelligent systems, Prentice Hall PTR, 1996.
6. Mesa Fernandez, J. M.; Cabal, V. A.; Montequin V. R.; Balsera, J. V.: Eng. Appl. Artif. Intel. 21 (2008), p. 1001.
7. Meradi, H.; Bouhouche, S.; Lahreche, M.: Eng. Technol. 24 (2008) No. 12, p. 851.
8. Rajesh, N.; Khare M. R.; Pabi, S. K.: Mater. Res. 13 (2010) No. 1, p. 15.
9. Staib, W. E.; Bills, N. G.; Staib. R. B.: Developments in neural network applications: the intelligent electric arc furnace, AISE Steel Technology, 1992, 69, No. 6, p. 29.
10. Stolte, G.: Secondary Metallurgy — Fundamentals Process Applications, Verlag Stahleisen GmbH, Düsseldorf, Germany, 2002.
11. Guanin, S.; Dimitrijevic, V.; Picciotto M.; Trassinelli, G.: Artificial Neural Network Approach for Molten Bath temperature Tracking on EAF/VD, Proc. 6^th Int. Conf. on Modelling and Simulation of Metallurgical Processes in Steelmaking (STEELSIM 2015), 23−25 September 2015, Bardolino, Italy.