Printer Friendly

Fuzzy system for implementing the cracks control during the continuous casting.


In (Tirian, 2008) is described the structure of the crack detection systems. Such cracks may occur during the continuous casting process, especially of those who are based on a neuronal network. This network makes a "1"-logical output signal when it detects a primary crack of the crystallizing apparatus; otherwise it produces a "0"-output signal. Any piece of information should be used properly in order to avoid any possible crack before the material exits the crystallizing apparatus (Adamy, 1999).

This paperwork comes up with an original fuzzy solution which could be added to the current structure of the control system of the continuous casting. Besides that, it uses all the features of the fuzzy logics (Lee, 1990) and it is able to predict any possible crack (Nakamura & Kazuho, 1996), providing with the best solutions and actions in order to prevent any cracks inside the crystallizing apparatus.

We could use this structure for any type of installation of continuous casting, but only along with the neuronal network for primary crack detection. Thus, considering the prediction principle we have chosen, we believe we are able to eliminate any fault during the casting process, when the cast material has cracks when coming out of the crystallizing apparatus.


Figure no. 1 describes the structure of the system we are suggesting. It is a non-linear non-dynamic fuzzy controller (Precup & Preitl, 1999) who has two different type of basis set on different rules. Fuzzy controller works based on two different rules bases on different rules. these rules are selected according to the signal they receive from the neuronal network in order to detect the primary cracks (Tirian, 2008):

a) Rule base "0" is still working though the neuronal network has not yet detected the primary cracks inside the crystallizing apparatus. We use four input features: the current casting speed, the primary water current flow, the temperature inside the distributor, and the technological risk. In such case, fuzzy controller is going to set the command for the casting speed and for the primary cooling water flow, in order to decrease the risk of crack occurrence.

It is for the first time when the crack occurrence risk is considered in case of the continuous casting, and specialized paperwork's refer to this matter.

b) Rule base "1" is working when the neuronal network detects some primary cracks of the material inside the crystallizing apparatus. In this case, the technological risk is negligible and the rest of the sizes could stay the same. We need a second rule base because although we set the same values of the input sizes, fuzzy controller must enable some much more obvious changes of the primary cooling water and of the casting speed than in the first case.

c) The technological risk (TR) contains a set of features established by the technical experts; they prove if they cracks may occur or not because of the chemical composition of the steel. Thus, a high percentage of Carbon (current 0.05-^0.17%) of the percentage between Mangane/Sulphurus, and so on, is an important feature who might cause some cracks. In order to set the exact values of the technological risk we have asked for the help from some experts in order to write down this paperwork. The mathematical model (Li, 2002) of the crust solidification process has allowed us to make some important conclusions referring to the technological risk of crack occurrence.

d) The two rules based shall be selected alternatively, according to the output values of the neuronal network for crack detection.

e) In order to come up with a rule base, we have to make some effort in order to analyze all possibilities. Therefore, we have contacted many experimented technical experts. In each case, we have analyzed thoroughly the crust solidification process using the mathematical model (Li, 2002). Of course, these rules based are not flawless, since they are influenced by a lot of factors who consider the state of the equipments and other practical features. Thus, we are going to improve their performance when being under stress and tested according to our new scheme. This method is usual in case of expert systems.




We have designed the fuzzy controller according to the "0" and "1" rule base. We have established the linguistic terms (for both input and output), the belonging functions, and the rule base.

These features have helped us find the command surfaces using the Matlab simulation.

In figure no. 2 we describe the block diagram of the fuzzy controller with rule base "0" value, meanwhile in the figure no. 3 we describe the block diagram rule base "1".


The two outputs of the fuzzy controller (0-1 values) shall be used for a flow calculation unit and for the appropriate speed, who work according to the relations:

[v.sub.c] = (1--[p.sub.v]) x [p.sup.*] (1)

[q.sub.c] = (1 + [p.sub.q]) x [q.sup.*] (2)

where [v.sup.*], [q.sup.*] are the required values for the casting speed and for the primary cooling water generated by the equipment.

Referring to the way this scheme works (figure no. 4), we should point out that:

a) As far as we can see, we have tried to avoid that any change should harm the already existing equipments; their features are only additional. All the regulating loop (they are situated inside the speed monitoring unit of the casting speed and of the primary cooling water parameters) and the expert systems work perfectly (Tirian et al., 2009). In this case, this is the main requirement, for economical and security purposes.

b) All the considerations from "point a" refer to several domains and we could use this scheme in order to disable any possible crack occurrence. This is the most appropriate solution for any type of equipment, involving low costs.

c) Input values of the fuzzy controller (q, v, t) are taken from the translating devices who already exist in all practical schemes. Usually, some of them are numerical translators, and other are analogical, but they are both highly precise. The technological risk (TR) is determined by the technical experts, according to the type of steel they use for the casting; it is introduced when the steel must be changed (usually, every other two days).



This paperwork introduces a new and original concept of the structure of a control system for the continuous casting. This new method avoids all the cracks of the cast material when it exits the crystallizing apparatus. The scheme comprises a fuzzy controller, based on a rule base, determined with the help of the mathematical model of the crust formation process. It also relies on the practical experience of the technical experts; it enables some change of the primary cooling water flow and of the casting speed. Implementation of the proposed scheme can be applied both on new installation and on an existence, without costly modifications.


Adamy, J. (1999). Device for early detection of run-out in continuous casting, United States Patent, No.5, 904,202, Date of Patent May 18th 1999

Lee, C. C. (1990). Fuzzy logic in control systems: Fuzzy logic controller, IEEE Trans. Systems, Man & Cybernetics 20(2): 4(04435

Li, C. (2002). Thermo-mechanical finite element model of shell behaviour in the continuous casting of steel, Sixth Asia-Pacific Symposium on Engineering Plasticity and its Applications, 2-6 December, Sydney, Australia, 2002

Nakamura T. & Kazuho K. (1996). Breakout prediction system in a continuous casting process, United States Patent, No.5, 548, 520, Date of Patent 20th August 1996

Precup, R.E & Preitl, St. (1999). Fuzzy Controllers, Editure Academic Horizons, pp 123-128, ISBN: 973-9400-61-2, Timisoara, Romania

Tirian G.O. (2008). Neural system for detecting cracks in the wire of the continuous casting, 12th International Research/Expert Conference--"Trends in the Development of Machinery and Associated Technology", August 26th 30th 2008, pp. 649-652 Istanbul, Turkey

Tirian, G.O., Anghel, S. & Pinca C. (2009). Control System of the Continuous Casting Process for Craks Removal, 5th International Symposium on Applied Computational Intelligence and Informatics, May 28-29, 2009, pp 265-269, ISBN:978-1-4244-4478-6, Timisoara, Romania
COPYRIGHT 2009 DAAAM International Vienna
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Tirian, Gelu Ovidiu; Prostean, Octavian; Rusu-Anghel, Stela; Pinca-Bretotean, Camelia; Cristea, Ana
Publication:Annals of DAAAM & Proceedings
Article Type:Report
Geographic Code:4EUAU
Date:Jan 1, 2009
Previous Article:Image retrieval based on color and shape.
Next Article:Optimization of deliming process.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |