Professional Documents
Culture Documents
Keywords: Pallet Car Tracking, Ladle Tracking, Torpedo Tracking, Diagnosis, Condition Monitoring, BF, EAF, Converter,
data based services
INTRODUCTION
Fully automatic detection of moving vessels like ladles, slagpots, torpedos, etc. is a common requirement in the steel
industry. Extensive heat, dust, smoke and changing lighting conditions are known problems which are complicating this task.
Primetals Technologies Austria GmbH has a new approach to overcome these problems. Basis is the three-dimensional
measurement of the vessel surface. The recognition and evaluation of characteristics and unique features leads to the
identification of the equipment.
DISCUSSION
Figure 2: raw distance data of a pallet car side wall with laser cut nameplate
In some cases the pixel information from 3D sensor might be invalid due to reflective surface or some other reasons. The
approach in this evaluation is to adjust the missing pixel value with respect to the pixel values of its neighbors.
Perspective corrections
Furthermore two surface planes were defined, one parallel to the sensor mounting position and one for the orientation of the
nameplate (figure 4, left). Using standard matrix rotation mathematics it is possible, to transform the distance values from the
nameplate plane to the sensor plane. The result is a normalized nameplate distance matrix showing the number in an
optimized reading angle shown in figure 4 (right.
Binarization
By applying standard thresholding to the image a ,a plane x’y’ which is parallel and very close to the nameplate plane could
be found. Therfore pixels over x’y’ represents the text and below represents the background (Figure 5).
Figure 9: weighing the holes according to their location provides the ladle number
Independent from the hole plate, the casting ladles at Dillinger are also providing numbers (Figure 10).
In order to be consistent with the actual trends of customized ladle identification for minimal adaption and vessel
modification also the identification of these numbers were examined.
Figure 11: raw sensor data at a distance of 2.2m (left) and pre calculated data (right)
The first evaluation step is compensating the ladle surface. For this calculation again the whole information of the vessel is
taken into consideration. The result including the 2cm thick number is shown in figure 11 (right).
The noise in the pre calculated data (see figure 11, right) could be reduced by applying standard filters. The result is shown in
Figure 12. The data is again provided to the number identification library which returns the identified number. Special
training of the OCR library was not necessary for identification of the numbers mounted on the ladles.
Figure 14: plugged (left) and unplugged (right) purging gas tubes
CONCLUSIONS
With test installations at voestalpine AG and Dillinger it was possible to show, that Primetals Technologies Austria GmbH
could provide a fully automated vessel and equipment tracking system based on TOF sensor technology.
The Tracking System can be utilized for all fully automated tracking tasks where temperature or changing light conditions are
limiting the use of RFID or optical identification systems.
The temperature of the equipment to be tracked is not affecting the sensor measurement data and subsequently the
identification. Because there is no electronic equipment mounted on the vessel to be tracked, the system could be used for
extreme high temperature vessel tracking applications.
If an equipment or vessel is already providing unique features, it is possible to use these features for identification. However,
it has to be taken into account that these identification features must fulfill certain criterias in order to be recognized by the
system.
REFERENCES
1. T. Kurzmann, F. Hartl, A. Rohrhofer, P. Aufreiter, ”Latest Innovations of Technological Automation Packages for
Steelmaking Plants” ESTAD 2015 Proceedings, Düsseldorf
2. A. Rohrhofer, T. Kurzmann, F. Hartl, P. Aufreiter, ”Advanced Automation Solution for Steelmaking and
Environmental Plants” AISTECH 2016 Proceedings, Pittsburgh
3. R. Smith, “An overview of the tesseract OCR Engine”, ICDAR 2007, Curitibar Brazil
4. Shi Zhixin, Setlur Srirangaraj and Govindaraju Venu, “Digital Image Enhancement Using Normalization Techniques
and their Application to Palm Leaf Manuscripts” CEDAR 2005, New York U.S.A.